Python 3
pip install pydicomCollecting pydicom Downloading https://files.pythonhosted.org/packages/f4/15/df16546bc59bfca390cf072d473fb2c8acd4231636f64356593a63137e55/pydicom-2.1.2-py3-none-any.whl (1.9MB) |████████████████████████████████| 1.9MB 4.7MB/s Installing collected packages: pydicom Successfully installed pydicom-2.1.2
%tensorflow_version 1.xTensorFlow 1.x selected.
xxxxxxxxxx!pip install keras==2.1.5Requirement already satisfied: keras==2.1.5 in /usr/local/lib/python3.7/dist-packages (2.1.5) Requirement already satisfied: pyyaml in /usr/local/lib/python3.7/dist-packages (from keras==2.1.5) (3.13) Requirement already satisfied: scipy>=0.14 in /usr/local/lib/python3.7/dist-packages (from keras==2.1.5) (1.4.1) Requirement already satisfied: six>=1.9.0 in /usr/local/lib/python3.7/dist-packages (from keras==2.1.5) (1.15.0) Requirement already satisfied: numpy>=1.9.1 in /usr/local/lib/python3.7/dist-packages (from keras==2.1.5) (1.19.5)
xxxxxxxxxximport osimport pandas as pd import numpy as npimport matplotlibimport matplotlib.pyplot as pltimport matplotlib.patches as patchesimport tqdmfrom tqdm import tqdm_notebookimport matplotlib.patches as patchesfrom matplotlib.patches import Rectangleimport seaborn as snsimport pydicom as dcmimport tensorflow as tffrom tensorflow import kerasimport osimport csvimport randomimport pydicomfrom skimage import iofrom skimage import measurefrom skimage.transform import resizefrom tensorflow.keras.applications.vgg19 import VGG19,preprocess_inputfrom tensorflow.keras.applications.inception_v3 import InceptionV3from tensorflow.keras.applications.mobilenet_v2 import MobileNetV2from tensorflow.keras.preprocessing.image import ImageDataGeneratorfrom tensorflow.keras.layers import Flatten,Dense,Dropout,BatchNormalization,LeakyReLU,ReLU,GaussianDropoutfrom tensorflow.keras.models import Modelfrom tensorflow.keras.callbacks import EarlyStopping,ReduceLROnPlateau%matplotlib inline PATH='/drive/MyDrive/GL_AIML/Capstone/'xxxxxxxxxx## Above we configure our notebook for googel colab, install/import necessary libraries and versionsxxxxxxxxxxclasses = pd.read_csv('/drive/MyDrive/GL_AIML/Capstone/stage_2_train_labels.csv')train = pd.read_csv('/drive/MyDrive/GL_AIML/Capstone/stage_2_detailed_class_info.csv') xxxxxxxxxxtrain| patientId | x | y | width | height | Target | |
|---|---|---|---|---|---|---|
| 0 | 0004cfab-14fd-4e49-80ba-63a80b6bddd6 | NaN | NaN | NaN | NaN | 0 |
| 1 | 00313ee0-9eaa-42f4-b0ab-c148ed3241cd | NaN | NaN | NaN | NaN | 0 |
| 2 | 00322d4d-1c29-4943-afc9-b6754be640eb | NaN | NaN | NaN | NaN | 0 |
| 3 | 003d8fa0-6bf1-40ed-b54c-ac657f8495c5 | NaN | NaN | NaN | NaN | 0 |
| 4 | 00436515-870c-4b36-a041-de91049b9ab4 | 264.0 | 152.0 | 213.0 | 379.0 | 1 |
| ... | ... | ... | ... | ... | ... | ... |
| 30222 | c1ec14ff-f6d7-4b38-b0cb-fe07041cbdc8 | 185.0 | 298.0 | 228.0 | 379.0 | 1 |
| 30223 | c1edf42b-5958-47ff-a1e7-4f23d99583ba | NaN | NaN | NaN | NaN | 0 |
| 30224 | c1f6b555-2eb1-4231-98f6-50a963976431 | NaN | NaN | NaN | NaN | 0 |
| 30225 | c1f7889a-9ea9-4acb-b64c-b737c929599a | 570.0 | 393.0 | 261.0 | 345.0 | 1 |
| 30226 | c1f7889a-9ea9-4acb-b64c-b737c929599a | 233.0 | 424.0 | 201.0 | 356.0 | 1 |
30227 rows × 6 columns
xxxxxxxxxx## Train has patient id's of patients along with bounding box coordinates of the opacity in the xray## We have a traget column which is binary and tells us if that particular patinet had pneumonia or notxxxxxxxxxxclasses| patientId | class | |
|---|---|---|
| 0 | 0004cfab-14fd-4e49-80ba-63a80b6bddd6 | No Lung Opacity / Not Normal |
| 1 | 00313ee0-9eaa-42f4-b0ab-c148ed3241cd | No Lung Opacity / Not Normal |
| 2 | 00322d4d-1c29-4943-afc9-b6754be640eb | No Lung Opacity / Not Normal |
| 3 | 003d8fa0-6bf1-40ed-b54c-ac657f8495c5 | Normal |
| 4 | 00436515-870c-4b36-a041-de91049b9ab4 | Lung Opacity |
| ... | ... | ... |
| 30222 | c1ec14ff-f6d7-4b38-b0cb-fe07041cbdc8 | Lung Opacity |
| 30223 | c1edf42b-5958-47ff-a1e7-4f23d99583ba | Normal |
| 30224 | c1f6b555-2eb1-4231-98f6-50a963976431 | Normal |
| 30225 | c1f7889a-9ea9-4acb-b64c-b737c929599a | Lung Opacity |
| 30226 | c1f7889a-9ea9-4acb-b64c-b737c929599a | Lung Opacity |
30227 rows × 2 columns
xxxxxxxxxx## Classes also has patient id's along with the class that it belongs toxxxxxxxxxxclasses['class'].unique()array(['No Lung Opacity / Not Normal', 'Normal', 'Lung Opacity'],
dtype=object)xxxxxxxxxx## Classes here are ### 1. No Lung Opacity / Not Normal### 2. Normal### 3. Lung Opacity## Here we get a baisc idea of whether the patient tested positive or negative for pneumoniaxxxxxxxxxxprint(f"Classes Summary-----> Data Points: {classes.shape[0]}, Classes: {classes.shape[1]}")print(f"Train Data Summary-----> Data Points: {train.shape[0]}, features: {train.shape[1]}")Classes Summary-----> Data Points: 30227, Classes: 2 Train Data Summary-----> Data Points: 30227, features: 6
xxxxxxxxxxdef missing_data(data): total = data.isnull().sum().sort_values(ascending = False) percent = (data.isnull().sum()/data.isnull().count()*100).sort_values(ascending = False) return np.transpose(pd.concat([total, percent], axis=1, keys=['Total', 'Percent']))xxxxxxxxxx## This above function will help us check the missing values and the % number compared to full dataxxxxxxxxxxdef get_feature_distribution(data, feature): label_counts = data[feature].value_counts() total_samples = len(data) print("Feature: {}".format(feature)) for i in range(len(label_counts)): label = label_counts.index[i] count = label_counts.values[i] percent = int((count / total_samples) * 10000) / 100 print("{:<30s}: {} or {}%".format(label, count, percent))xxxxxxxxxx## This function gives us the % of data in each class against the totalxxxxxxxxxxdef show_dicom_images(data): img_data = list(data.T.to_dict().values()) f, ax = plt.subplots(3,3, figsize=(16,18)) for i,data_row in enumerate(img_data): patientImage = data_row['patientId']+'.dcm' imagePath = os.path.join('stage_2_train_images',patientImage) data_row_img_data = dcm.read_file(imagePath) modality = data_row_img_data.Modality age = data_row_img_data.PatientAge sex = data_row_img_data.PatientSex data_row_img = dcm.dcmread(imagePath) ax[i//3, i%3].imshow(data_row_img.pixel_array, cmap=plt.cm.bone) ax[i//3, i%3].axis('off') ax[i//3, i%3].set_title('ID: {}\nModality: {} Age: {} Sex: {} Target: {}\nClass: {}\nWindow: {}:{}:{}:{}'.format( data_row['patientId'], modality, age, sex, data_row['Target'], data_row['class'], data_row['x'],data_row['y'],data_row['width'],data_row['height'])) plt.show()xxxxxxxxxx## The above function helps us explore and view our images along with the dicom files. Medical images are preserved in peculiar format called DICOM files with extension *.dcm . It has combination of header metadata and underlying raw image array. Python has support for this and the best module to be used is the pydicom modulexxxxxxxxxxdef show_dicom_images_with_boxes(data): img_data = list(data.T.to_dict().values()) f, ax = plt.subplots(3,3, figsize=(16,18)) for i,data_row in enumerate(img_data): patientImage = data_row['patientId']+'.dcm' imagePath = os.path.join('stage_2_train_images',patientImage) data_row_img_data = dcm.read_file(imagePath) modality = data_row_img_data.Modality age = data_row_img_data.PatientAge sex = data_row_img_data.PatientSex data_row_img = dcm.dcmread(imagePath) ax[i//3, i%3].imshow(data_row_img.pixel_array, cmap=plt.cm.bone) ax[i//3, i%3].axis('off') ax[i//3, i%3].set_title('ID: {}\nModality: {} Age: {} Sex: {} Target: {}\nClass: {}'.format( data_row['patientId'],modality, age, sex, data_row['Target'], data_row['class'])) rows = train[train['patientId']==data_row['patientId']] box_data = list(rows.T.to_dict().values()) for j, row in enumerate(box_data): ax[i//3, i%3].add_patch(Rectangle(xy=(row['x'], row['y']), width=row['width'],height=row['height'], color="yellow",alpha = 0.1)) plt.show()xxxxxxxxxxmissing_data(train)| height | width | y | x | Target | patientId | |
|---|---|---|---|---|---|---|
| Total | 20672.000000 | 20672.000000 | 20672.000000 | 20672.000000 | 0.0 | 0.0 |
| Percent | 68.389188 | 68.389188 | 68.389188 | 68.389188 | 0.0 | 0.0 |
xxxxxxxxxx## We check missing values in our train dataframe## There are an ample amount in co ordinates since xray images with no Pneumonia needs no bouding boxes## Hence H,W,X,W would be nan for class 0## We shall not change it and let it be, because these will be handy further, since we can easily separte classes by filtering with np.nanxxxxxxxxxxmissing_data(classes)| class | patientId | |
|---|---|---|
| Total | 0.0 | 0.0 |
| Percent | 0.0 | 0.0 |
xxxxxxxxxx## The classes dataframe is free of null values xxxxxxxxxxf, ax = plt.subplots(1,1, figsize=(15,10))total = float(len(classes))sns.countplot(classes['class'],order = classes['class'].value_counts().index, palette='Set3')for p in ax.patches: height = p.get_height() ax.text(p.get_x()+p.get_width()/2., height + 3, '{:1.2f}%'.format(100*height/total), ha="center") plt.show()/Users/chiraglodaya/opt/anaconda3/lib/python3.8/site-packages/seaborn/_decorators.py:36: FutureWarning: Pass the following variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation. warnings.warn(
xxxxxxxxxxget_feature_distribution(classes, 'class')Feature: class No Lung Opacity / Not Normal : 11821 or 39.1% Lung Opacity : 9555 or 31.61% Normal : 8851 or 29.28%
xxxxxxxxxx## Above we see our target class distribution## No Lung Opacity / Not Normal is ~ 39% which is negative class along with another negative class Normal which is ~30%## Our positive class Lung Opacity is ~31%xxxxxxxxxxtrain2 = train.merge(classes, left_on='patientId', right_on='patientId', how='inner')train2| patientId | x | y | width | height | Target | class | |
|---|---|---|---|---|---|---|---|
| 0 | 0004cfab-14fd-4e49-80ba-63a80b6bddd6 | NaN | NaN | NaN | NaN | 0 | No Lung Opacity / Not Normal |
| 1 | 00313ee0-9eaa-42f4-b0ab-c148ed3241cd | NaN | NaN | NaN | NaN | 0 | No Lung Opacity / Not Normal |
| 2 | 00322d4d-1c29-4943-afc9-b6754be640eb | NaN | NaN | NaN | NaN | 0 | No Lung Opacity / Not Normal |
| 3 | 003d8fa0-6bf1-40ed-b54c-ac657f8495c5 | NaN | NaN | NaN | NaN | 0 | Normal |
| 4 | 00436515-870c-4b36-a041-de91049b9ab4 | 264.0 | 152.0 | 213.0 | 379.0 | 1 | Lung Opacity |
| ... | ... | ... | ... | ... | ... | ... | ... |
| 37624 | c1f6b555-2eb1-4231-98f6-50a963976431 | NaN | NaN | NaN | NaN | 0 | Normal |
| 37625 | c1f7889a-9ea9-4acb-b64c-b737c929599a | 570.0 | 393.0 | 261.0 | 345.0 | 1 | Lung Opacity |
| 37626 | c1f7889a-9ea9-4acb-b64c-b737c929599a | 570.0 | 393.0 | 261.0 | 345.0 | 1 | Lung Opacity |
| 37627 | c1f7889a-9ea9-4acb-b64c-b737c929599a | 233.0 | 424.0 | 201.0 | 356.0 | 1 | Lung Opacity |
| 37628 | c1f7889a-9ea9-4acb-b64c-b737c929599a | 233.0 | 424.0 | 201.0 | 356.0 | 1 | Lung Opacity |
37629 rows × 7 columns
xxxxxxxxxx## We next merge the datafarmes into a single train dataframe for better usgaexxxxxxxxxxfig, ax = plt.subplots(nrows=1,figsize=(15,10))tmp = train2.groupby('Target')['class'].value_counts()df = pd.DataFrame(data={'Numbers': tmp.values}, index=tmp.index).reset_index()sns.barplot(ax=ax,x = 'Target', y='Numbers',hue='class',data=df, palette='Set3')plt.title("Target Classes")plt.show()
xxxxxxxxxx## We now check overall distribution of our classes positive and negative## Positive is almost 69% and negative class is 31%## We get an idea that class imbalance is not a case herexxxxxxxxxxtarget1 = train2[train2['Target']==1]sns.set_style('whitegrid')plt.figure()fig, ax = plt.subplots(2,2,figsize=(20,20))sns.distplot(target1['x'],kde=True,bins=50, color="red", ax=ax[0,0])sns.distplot(target1['y'],kde=True,bins=50, color="blue", ax=ax[0,1])sns.distplot(target1['width'],kde=True,bins=50, color="green", ax=ax[1,0])sns.distplot(target1['height'],kde=True,bins=50, color="magenta", ax=ax[1,1])locs, labels = plt.xticks()plt.tick_params(axis='both', which='major', labelsize=12)plt.show()/Users/chiraglodaya/opt/anaconda3/lib/python3.8/site-packages/seaborn/distributions.py:2551: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms). warnings.warn(msg, FutureWarning) /Users/chiraglodaya/opt/anaconda3/lib/python3.8/site-packages/seaborn/distributions.py:2551: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms). warnings.warn(msg, FutureWarning) /Users/chiraglodaya/opt/anaconda3/lib/python3.8/site-packages/seaborn/distributions.py:2551: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms). warnings.warn(msg, FutureWarning) /Users/chiraglodaya/opt/anaconda3/lib/python3.8/site-packages/seaborn/distributions.py:2551: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms). warnings.warn(msg, FutureWarning)
<Figure size 432x288 with 0 Axes>
xxxxxxxxxx## We check the distribution of our x,y,h,w## Y, H, W seem to be normally distributed but X seems worrysome with bimodal disrtibution which may cause problems furtherxxxxxxxxxxfig, ax = plt.subplots(1,1,figsize=(7,7))target_sample = target1.sample(2000)target_sample['xc'] = target_sample['x'] + target_sample['width'] / 2target_sample['yc'] = target_sample['y'] + target_sample['height'] / 2plt.title("Centers of Lung Opacity rectangles (brown) over rectangles (blue)\nSample size: 2000")target_sample.plot.scatter(x='xc', y='yc', xlim=(0,1024), ylim=(0,1024), ax=ax, alpha=0.8, marker=".", color="black")for i, crt_sample in target_sample.iterrows(): ax.add_patch(Rectangle(xy=(crt_sample['x'], crt_sample['y']), width=crt_sample['width'],height=crt_sample['height'],alpha=3.5e-3, color="red"))plt.show()
xxxxxxxxxx## We next consider 2000 samples of pneumonia data and jot the center of the binding boxes## The red patches are the entire boxes and black points are the center of those boxes## With this we get an very important idea that no specific part of the lung may be prone to opacity## Almost we have every part of lung in 2000 samples has opacityxxxxxxxxxximage_train_path = os.listdir('stage_2_train_images')image_test_path = os.listdir('stage_2_test_images')print("Number of images in train set:", len(image_train_path),"\nNumber of images in test set:", len(image_test_path))Number of images in train set: 26684 Number of images in test set: 3000
xxxxxxxxxx## So we check the train and test set, train set excluding repeated images is 26684 and test has 3000 samples## This much data is good to train a Deep learning model from scratch with decent accuracy and more than enough to feed pre trained modelxxxxxxxxxxprint("Unique patientId in training set: ", train2['patientId'].nunique())Unique patientId in training set: 26684
xxxxxxxxxxsamplePatientID = list(train2[:3].T.to_dict().values())[0]['patientId']samplePatientID = samplePatientID+'.dcm'dicom_file_path = os.path.join('stage_2_train_images',samplePatientID)dicom_file_dataset = dcm.read_file(dicom_file_path)dicom_file_datasetDataset.file_meta ------------------------------- (0002, 0000) File Meta Information Group Length UL: 202 (0002, 0001) File Meta Information Version OB: b'\x00\x01' (0002, 0002) Media Storage SOP Class UID UI: Secondary Capture Image Storage (0002, 0003) Media Storage SOP Instance UID UI: 1.2.276.0.7230010.3.1.4.8323329.28530.1517874485.775526 (0002, 0010) Transfer Syntax UID UI: JPEG Baseline (Process 1) (0002, 0012) Implementation Class UID UI: 1.2.276.0.7230010.3.0.3.6.0 (0002, 0013) Implementation Version Name SH: 'OFFIS_DCMTK_360' ------------------------------------------------- (0008, 0005) Specific Character Set CS: 'ISO_IR 100' (0008, 0016) SOP Class UID UI: Secondary Capture Image Storage (0008, 0018) SOP Instance UID UI: 1.2.276.0.7230010.3.1.4.8323329.28530.1517874485.775526 (0008, 0020) Study Date DA: '19010101' (0008, 0030) Study Time TM: '000000.00' (0008, 0050) Accession Number SH: '' (0008, 0060) Modality CS: 'CR' (0008, 0064) Conversion Type CS: 'WSD' (0008, 0090) Referring Physician's Name PN: '' (0008, 103e) Series Description LO: 'view: PA' (0010, 0010) Patient's Name PN: '0004cfab-14fd-4e49-80ba-63a80b6bddd6' (0010, 0020) Patient ID LO: '0004cfab-14fd-4e49-80ba-63a80b6bddd6' (0010, 0030) Patient's Birth Date DA: '' (0010, 0040) Patient's Sex CS: 'F' (0010, 1010) Patient's Age AS: '51' (0018, 0015) Body Part Examined CS: 'CHEST' (0018, 5101) View Position CS: 'PA' (0020, 000d) Study Instance UID UI: 1.2.276.0.7230010.3.1.2.8323329.28530.1517874485.775525 (0020, 000e) Series Instance UID UI: 1.2.276.0.7230010.3.1.3.8323329.28530.1517874485.775524 (0020, 0010) Study ID SH: '' (0020, 0011) Series Number IS: "1" (0020, 0013) Instance Number IS: "1" (0020, 0020) Patient Orientation CS: '' (0028, 0002) Samples per Pixel US: 1 (0028, 0004) Photometric Interpretation CS: 'MONOCHROME2' (0028, 0010) Rows US: 1024 (0028, 0011) Columns US: 1024 (0028, 0030) Pixel Spacing DS: [0.14300000000000002, 0.14300000000000002] (0028, 0100) Bits Allocated US: 8 (0028, 0101) Bits Stored US: 8 (0028, 0102) High Bit US: 7 (0028, 0103) Pixel Representation US: 0 (0028, 2110) Lossy Image Compression CS: '01' (0028, 2114) Lossy Image Compression Method CS: 'ISO_10918_1' (7fe0, 0010) Pixel Data OB: Array of 142006 elements
xxxxxxxxxx## Above we check how dicom metadat looks likexxxxxxxxxxpatientId = train2['patientId'][0]dcm_file = f'stage_2_train_images/{patientId}.dcm'dcm_data = dcm.read_file(dcm_file)im = dcm_data.pixel_arrayprint(type(im))print(im.dtype)print(im.shape)<class 'numpy.ndarray'> uint8 (1024, 1024)
xxxxxxxxxx## Also we see that above, our images are numpy arrays with dimension 1024x1024## These dimensions are perfect for training model as they are AI friendly, not extremely high dimensiona nor extremely pixalated## This saves us from image transformation or any sort of data compressionxxxxxxxxxxshow_dicom_images(train2[train2['Target']==1].sample(9))
xxxxxxxxxxshow_dicom_images_with_boxes(train2[train2['Target']==1].sample(9))
xxxxxxxxxxshow_dicom_images(train2[train2['Target']==0].sample(9))
xxxxxxxxxx## Above we visualise the images along with basic dicom metadat## We see positive class as well as negative class and also with binding boxes## One thing that is evident above is the opacity is not visually identifyablexxxxxxxxxxtrain_labels_df = pd.read_csv('/drive/MyDrive/GL_AIML/Capstone/stage_2_train_labels.csv')class_info_df = pd.read_csv('/drive/MyDrive/GL_AIML/Capstone/stage_2_detailed_class_info.csv')xxxxxxxxxxTRAIN_PATH = "/drive/MyDrive/GL_AIML/Capstone/stage_2_train_images"TEST_PATH = "/drive/MyDrive/GL_AIML/Capstone/stage_2_test_images"train_class_df = train_labels_df.merge(class_info_df, left_on='patientId', right_on='patientId', how='inner')image_sample_path = os.listdir(PATH+'/stage_2_train_images')[:5]image_train_path = os.listdir(PATH+'/stage_2_train_images')image_test_path = os.listdir(PATH+'/stage_2_test_images')xxxxxxxxxxpneumonia_locations = {}with open(os.path.join(PATH+'/stage_2_train_labels.csv'), mode='r') as infile: reader = csv.reader(infile) next(reader, None) for rows in reader: filename = rows[0] location = rows[1:5] pneumonia = rows[5] if pneumonia == '1': location = [int(float(i)) for i in location] if filename in pneumonia_locations: pneumonia_locations[filename].append(location) else: pneumonia_locations[filename] = [location]xxxxxxxxxx## Next we make a block of code to get all the positive samples into a dictionary.xxxxxxxxxxfolder = PATH+'/stage_2_train_images'filenames = os.listdir(folder)random.shuffle(filenames)n_valid_samples = 2560train_filenames = filenames[n_valid_samples:]valid_filenames = filenames[:n_valid_samples]print('n train samples', len(train_filenames))print('n valid samples', len(valid_filenames))n_train_samples = len(filenames) - n_valid_samplesprint('Total train images:',len(filenames))print('Images with pneumonia:', len(pneumonia_locations))n train samples 24154 n valid samples 2560 Total train images: 26714 Images with pneumonia: 6012
xxxxxxxxxxclass generator(keras.utils.Sequence): def __init__(self, folder, filenames, pneumonia_locations=None, batch_size=32, image_size=256, shuffle=True, augment=False, predict=False): self.folder = folder self.filenames = filenames self.pneumonia_locations = pneumonia_locations self.batch_size = batch_size self.image_size = image_size self.shuffle = shuffle self.augment = augment self.predict = predict self.on_epoch_end() def __load__(self, filename): img = pydicom.dcmread(os.path.join(self.folder, filename)).pixel_array msk = np.zeros(img.shape) if filename in pneumonia_locations: for location in pneumonia_locations[filename]: x, y, w, h = location msk[y:y+h, x:x+w] = 1 if self.augment and random.random() > 0.5: img = np.fliplr(img) msk = np.fliplr(msk) img = resize(img, (self.image_size, self.image_size), mode='symmetric') msk = resize(msk, (self.image_size, self.image_size), mode='symmetric') > 0.5 img = np.expand_dims(img, -1) msk = np.expand_dims(msk, -1) return img, msk def __loadpredict__(self, filename): img = pydicom.dcmread(os.path.join(self.folder, filename)).pixel_array img = resize(img, (self.image_size, self.image_size), mode='symmetric') img = np.expand_dims(img, -1) return img def __getitem__(self, index): filenames = self.filenames[index*self.batch_size:(index+1)*self.batch_size] if self.predict: imgs = [self.__loadpredict__(filename) for filename in filenames] imgs = np.array(imgs) return imgs, filenames else: items = [self.__load__(filename) for filename in filenames] imgs, msks = zip(*items) imgs = np.array(imgs) msks = np.array(msks) return imgs, msks def on_epoch_end(self): if self.shuffle: random.shuffle(self.filenames) def __len__(self): if self.predict: return int(np.ceil(len(self.filenames) / self.batch_size)) else: return int(len(self.filenames) / self.batch_size)xxxxxxxxxx## Next we make a generator class whihc has load data, get data and predict data functionsxxxxxxxxxxdef create_downsample(channels, inputs): x = keras.layers.BatchNormalization(momentum=0.9)(inputs) x = keras.layers.LeakyReLU(0)(x) x = keras.layers.Conv2D(channels, 1, padding='same', use_bias=False)(x) x = keras.layers.MaxPool2D(2)(x) return xdef create_resblock(channels, inputs): x = keras.layers.BatchNormalization(momentum=0.9)(inputs) x = keras.layers.LeakyReLU(0)(x) x = keras.layers.Conv2D(channels, 3, padding='same', use_bias=False)(x) x = keras.layers.BatchNormalization(momentum=0.9)(x) x = keras.layers.LeakyReLU(0)(x) x = keras.layers.Conv2D(channels, 3, padding='same', use_bias=False)(x) x = keras.layers.BatchNormalization(momentum=0.9)(x) x = keras.layers.LeakyReLU(0)(x) x = keras.layers.Conv2D(channels, 3, padding='same', use_bias=False)(x) addInput = x; print("Add input shape:", addInput.shape) print("Resnet block input shape:", inputs.shape) resBlockOut = keras.layers.add([addInput, inputs]) print("Resnet block out shape:", resBlockOut.shape) out = keras.layers.concatenate([resBlockOut, addInput], axis=3) print("concat block out shape:", out.shape) out = keras.layers.Conv2D(channels, 1, padding='same', use_bias=False)(out) print("mixed block out shape:", out.shape) return outdef create_network(input_size, channels, n_blocks=2, depth=4): inputs = keras.Input(shape=(input_size, input_size, 1)) x = keras.layers.Conv2D(channels, 3, padding='same', use_bias=False)(inputs) for d in range(depth): channels = channels * 2 x = create_downsample(channels, x) for b in range(n_blocks): x = create_resblock(channels, x) x = keras.layers.BatchNormalization(momentum=0.9)(x) x = keras.layers.LeakyReLU(0)(x) x = keras.layers.Conv2D(1, 1, activation='sigmoid')(x) outputs = keras.layers.UpSampling2D(2**depth)(x) model = keras.Model(inputs=inputs, outputs=outputs) return modelxxxxxxxxxx## Next we write functions to downsample the data(Since positive class is very less than negative class), create resnet layers and complete the network.xxxxxxxxxxdef iou_loss(y_true, y_pred): y_true = tf.reshape(y_true, [-1]) y_pred = tf.reshape(y_pred, [-1]) intersection = tf.reduce_sum(y_true * y_pred) score = (intersection + 1.) / (tf.reduce_sum(y_true) + tf.reduce_sum(y_pred) - intersection + 1.) return 1 - scoredef iou_bce_loss(y_true, y_pred): return 0.5 * keras.losses.binary_crossentropy(y_true, y_pred) + 0.5 * iou_loss(y_true, y_pred)def mean_iou(y_true, y_pred): y_pred = tf.round(y_pred) intersect = tf.reduce_sum(y_true * y_pred, axis=[1, 2, 3]) union = tf.reduce_sum(y_true, axis=[1, 2, 3]) + tf.reduce_sum(y_pred, axis=[1, 2, 3]) smooth = tf.ones(tf.shape(intersect)) return tf.reduce_mean((intersect + smooth) / (union - intersect + smooth))xxxxxxxxxx## Next we write functions to get jaccord loss metricxxxxxxxxxxmodel = create_network(input_size=128, channels=16, n_blocks=2, depth=3)model.compile(optimizer='adam', loss=iou_bce_loss, metrics=['accuracy', mean_iou])print("model summary:", model.summary())def cosine_annealing(x): lr = 0.001 epochs = 25 return lr*(np.cos(np.pi*x/epochs)+1.)/2learning_rate = tf.keras.callbacks.LearningRateScheduler(cosine_annealing)WARNING:tensorflow:From /tensorflow-1.15.2/python3.7/tensorflow_core/python/ops/resource_variable_ops.py:1630: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.
Instructions for updating:
If using Keras pass *_constraint arguments to layers.
Add input shape: (?, 64, 64, 32)
Resnet block input shape: (?, 64, 64, 32)
Resnet block out shape: (?, 64, 64, 32)
concat block out shape: (?, 64, 64, 64)
mixed block out shape: (?, 64, 64, 32)
Add input shape: (?, 64, 64, 32)
Resnet block input shape: (?, 64, 64, 32)
Resnet block out shape: (?, 64, 64, 32)
concat block out shape: (?, 64, 64, 64)
mixed block out shape: (?, 64, 64, 32)
Add input shape: (?, 32, 32, 64)
Resnet block input shape: (?, 32, 32, 64)
Resnet block out shape: (?, 32, 32, 64)
concat block out shape: (?, 32, 32, 128)
mixed block out shape: (?, 32, 32, 64)
Add input shape: (?, 32, 32, 64)
Resnet block input shape: (?, 32, 32, 64)
Resnet block out shape: (?, 32, 32, 64)
concat block out shape: (?, 32, 32, 128)
mixed block out shape: (?, 32, 32, 64)
Add input shape: (?, 16, 16, 128)
Resnet block input shape: (?, 16, 16, 128)
Resnet block out shape: (?, 16, 16, 128)
concat block out shape: (?, 16, 16, 256)
mixed block out shape: (?, 16, 16, 128)
Add input shape: (?, 16, 16, 128)
Resnet block input shape: (?, 16, 16, 128)
Resnet block out shape: (?, 16, 16, 128)
concat block out shape: (?, 16, 16, 256)
mixed block out shape: (?, 16, 16, 128)
Model: "model"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_1 (InputLayer) [(None, 128, 128, 1) 0
__________________________________________________________________________________________________
conv2d (Conv2D) (None, 128, 128, 16) 144 input_1[0][0]
__________________________________________________________________________________________________
batch_normalization (BatchNorma (None, 128, 128, 16) 64 conv2d[0][0]
__________________________________________________________________________________________________
leaky_re_lu (LeakyReLU) (None, 128, 128, 16) 0 batch_normalization[0][0]
__________________________________________________________________________________________________
conv2d_1 (Conv2D) (None, 128, 128, 32) 512 leaky_re_lu[0][0]
__________________________________________________________________________________________________
max_pooling2d (MaxPooling2D) (None, 64, 64, 32) 0 conv2d_1[0][0]
__________________________________________________________________________________________________
batch_normalization_1 (BatchNor (None, 64, 64, 32) 128 max_pooling2d[0][0]
__________________________________________________________________________________________________
leaky_re_lu_1 (LeakyReLU) (None, 64, 64, 32) 0 batch_normalization_1[0][0]
__________________________________________________________________________________________________
conv2d_2 (Conv2D) (None, 64, 64, 32) 9216 leaky_re_lu_1[0][0]
__________________________________________________________________________________________________
batch_normalization_2 (BatchNor (None, 64, 64, 32) 128 conv2d_2[0][0]
__________________________________________________________________________________________________
leaky_re_lu_2 (LeakyReLU) (None, 64, 64, 32) 0 batch_normalization_2[0][0]
__________________________________________________________________________________________________
conv2d_3 (Conv2D) (None, 64, 64, 32) 9216 leaky_re_lu_2[0][0]
__________________________________________________________________________________________________
batch_normalization_3 (BatchNor (None, 64, 64, 32) 128 conv2d_3[0][0]
__________________________________________________________________________________________________
leaky_re_lu_3 (LeakyReLU) (None, 64, 64, 32) 0 batch_normalization_3[0][0]
__________________________________________________________________________________________________
conv2d_4 (Conv2D) (None, 64, 64, 32) 9216 leaky_re_lu_3[0][0]
__________________________________________________________________________________________________
add (Add) (None, 64, 64, 32) 0 conv2d_4[0][0]
max_pooling2d[0][0]
__________________________________________________________________________________________________
concatenate (Concatenate) (None, 64, 64, 64) 0 add[0][0]
conv2d_4[0][0]
__________________________________________________________________________________________________
conv2d_5 (Conv2D) (None, 64, 64, 32) 2048 concatenate[0][0]
__________________________________________________________________________________________________
batch_normalization_4 (BatchNor (None, 64, 64, 32) 128 conv2d_5[0][0]
__________________________________________________________________________________________________
leaky_re_lu_4 (LeakyReLU) (None, 64, 64, 32) 0 batch_normalization_4[0][0]
__________________________________________________________________________________________________
conv2d_6 (Conv2D) (None, 64, 64, 32) 9216 leaky_re_lu_4[0][0]
__________________________________________________________________________________________________
batch_normalization_5 (BatchNor (None, 64, 64, 32) 128 conv2d_6[0][0]
__________________________________________________________________________________________________
leaky_re_lu_5 (LeakyReLU) (None, 64, 64, 32) 0 batch_normalization_5[0][0]
__________________________________________________________________________________________________
conv2d_7 (Conv2D) (None, 64, 64, 32) 9216 leaky_re_lu_5[0][0]
__________________________________________________________________________________________________
batch_normalization_6 (BatchNor (None, 64, 64, 32) 128 conv2d_7[0][0]
__________________________________________________________________________________________________
leaky_re_lu_6 (LeakyReLU) (None, 64, 64, 32) 0 batch_normalization_6[0][0]
__________________________________________________________________________________________________
conv2d_8 (Conv2D) (None, 64, 64, 32) 9216 leaky_re_lu_6[0][0]
__________________________________________________________________________________________________
add_1 (Add) (None, 64, 64, 32) 0 conv2d_8[0][0]
conv2d_5[0][0]
__________________________________________________________________________________________________
concatenate_1 (Concatenate) (None, 64, 64, 64) 0 add_1[0][0]
conv2d_8[0][0]
__________________________________________________________________________________________________
conv2d_9 (Conv2D) (None, 64, 64, 32) 2048 concatenate_1[0][0]
__________________________________________________________________________________________________
batch_normalization_7 (BatchNor (None, 64, 64, 32) 128 conv2d_9[0][0]
__________________________________________________________________________________________________
leaky_re_lu_7 (LeakyReLU) (None, 64, 64, 32) 0 batch_normalization_7[0][0]
__________________________________________________________________________________________________
conv2d_10 (Conv2D) (None, 64, 64, 64) 2048 leaky_re_lu_7[0][0]
__________________________________________________________________________________________________
max_pooling2d_1 (MaxPooling2D) (None, 32, 32, 64) 0 conv2d_10[0][0]
__________________________________________________________________________________________________
batch_normalization_8 (BatchNor (None, 32, 32, 64) 256 max_pooling2d_1[0][0]
__________________________________________________________________________________________________
leaky_re_lu_8 (LeakyReLU) (None, 32, 32, 64) 0 batch_normalization_8[0][0]
__________________________________________________________________________________________________
conv2d_11 (Conv2D) (None, 32, 32, 64) 36864 leaky_re_lu_8[0][0]
__________________________________________________________________________________________________
batch_normalization_9 (BatchNor (None, 32, 32, 64) 256 conv2d_11[0][0]
__________________________________________________________________________________________________
leaky_re_lu_9 (LeakyReLU) (None, 32, 32, 64) 0 batch_normalization_9[0][0]
__________________________________________________________________________________________________
conv2d_12 (Conv2D) (None, 32, 32, 64) 36864 leaky_re_lu_9[0][0]
__________________________________________________________________________________________________
batch_normalization_10 (BatchNo (None, 32, 32, 64) 256 conv2d_12[0][0]
__________________________________________________________________________________________________
leaky_re_lu_10 (LeakyReLU) (None, 32, 32, 64) 0 batch_normalization_10[0][0]
__________________________________________________________________________________________________
conv2d_13 (Conv2D) (None, 32, 32, 64) 36864 leaky_re_lu_10[0][0]
__________________________________________________________________________________________________
add_2 (Add) (None, 32, 32, 64) 0 conv2d_13[0][0]
max_pooling2d_1[0][0]
__________________________________________________________________________________________________
concatenate_2 (Concatenate) (None, 32, 32, 128) 0 add_2[0][0]
conv2d_13[0][0]
__________________________________________________________________________________________________
conv2d_14 (Conv2D) (None, 32, 32, 64) 8192 concatenate_2[0][0]
__________________________________________________________________________________________________
batch_normalization_11 (BatchNo (None, 32, 32, 64) 256 conv2d_14[0][0]
__________________________________________________________________________________________________
leaky_re_lu_11 (LeakyReLU) (None, 32, 32, 64) 0 batch_normalization_11[0][0]
__________________________________________________________________________________________________
conv2d_15 (Conv2D) (None, 32, 32, 64) 36864 leaky_re_lu_11[0][0]
__________________________________________________________________________________________________
batch_normalization_12 (BatchNo (None, 32, 32, 64) 256 conv2d_15[0][0]
__________________________________________________________________________________________________
leaky_re_lu_12 (LeakyReLU) (None, 32, 32, 64) 0 batch_normalization_12[0][0]
__________________________________________________________________________________________________
conv2d_16 (Conv2D) (None, 32, 32, 64) 36864 leaky_re_lu_12[0][0]
__________________________________________________________________________________________________
batch_normalization_13 (BatchNo (None, 32, 32, 64) 256 conv2d_16[0][0]
__________________________________________________________________________________________________
leaky_re_lu_13 (LeakyReLU) (None, 32, 32, 64) 0 batch_normalization_13[0][0]
__________________________________________________________________________________________________
conv2d_17 (Conv2D) (None, 32, 32, 64) 36864 leaky_re_lu_13[0][0]
__________________________________________________________________________________________________
add_3 (Add) (None, 32, 32, 64) 0 conv2d_17[0][0]
conv2d_14[0][0]
__________________________________________________________________________________________________
concatenate_3 (Concatenate) (None, 32, 32, 128) 0 add_3[0][0]
conv2d_17[0][0]
__________________________________________________________________________________________________
conv2d_18 (Conv2D) (None, 32, 32, 64) 8192 concatenate_3[0][0]
__________________________________________________________________________________________________
batch_normalization_14 (BatchNo (None, 32, 32, 64) 256 conv2d_18[0][0]
__________________________________________________________________________________________________
leaky_re_lu_14 (LeakyReLU) (None, 32, 32, 64) 0 batch_normalization_14[0][0]
__________________________________________________________________________________________________
conv2d_19 (Conv2D) (None, 32, 32, 128) 8192 leaky_re_lu_14[0][0]
__________________________________________________________________________________________________
max_pooling2d_2 (MaxPooling2D) (None, 16, 16, 128) 0 conv2d_19[0][0]
__________________________________________________________________________________________________
batch_normalization_15 (BatchNo (None, 16, 16, 128) 512 max_pooling2d_2[0][0]
__________________________________________________________________________________________________
leaky_re_lu_15 (LeakyReLU) (None, 16, 16, 128) 0 batch_normalization_15[0][0]
__________________________________________________________________________________________________
conv2d_20 (Conv2D) (None, 16, 16, 128) 147456 leaky_re_lu_15[0][0]
__________________________________________________________________________________________________
batch_normalization_16 (BatchNo (None, 16, 16, 128) 512 conv2d_20[0][0]
__________________________________________________________________________________________________
leaky_re_lu_16 (LeakyReLU) (None, 16, 16, 128) 0 batch_normalization_16[0][0]
__________________________________________________________________________________________________
conv2d_21 (Conv2D) (None, 16, 16, 128) 147456 leaky_re_lu_16[0][0]
__________________________________________________________________________________________________
batch_normalization_17 (BatchNo (None, 16, 16, 128) 512 conv2d_21[0][0]
__________________________________________________________________________________________________
leaky_re_lu_17 (LeakyReLU) (None, 16, 16, 128) 0 batch_normalization_17[0][0]
__________________________________________________________________________________________________
conv2d_22 (Conv2D) (None, 16, 16, 128) 147456 leaky_re_lu_17[0][0]
__________________________________________________________________________________________________
add_4 (Add) (None, 16, 16, 128) 0 conv2d_22[0][0]
max_pooling2d_2[0][0]
__________________________________________________________________________________________________
concatenate_4 (Concatenate) (None, 16, 16, 256) 0 add_4[0][0]
conv2d_22[0][0]
__________________________________________________________________________________________________
conv2d_23 (Conv2D) (None, 16, 16, 128) 32768 concatenate_4[0][0]
__________________________________________________________________________________________________
batch_normalization_18 (BatchNo (None, 16, 16, 128) 512 conv2d_23[0][0]
__________________________________________________________________________________________________
leaky_re_lu_18 (LeakyReLU) (None, 16, 16, 128) 0 batch_normalization_18[0][0]
__________________________________________________________________________________________________
conv2d_24 (Conv2D) (None, 16, 16, 128) 147456 leaky_re_lu_18[0][0]
__________________________________________________________________________________________________
batch_normalization_19 (BatchNo (None, 16, 16, 128) 512 conv2d_24[0][0]
__________________________________________________________________________________________________
leaky_re_lu_19 (LeakyReLU) (None, 16, 16, 128) 0 batch_normalization_19[0][0]
__________________________________________________________________________________________________
conv2d_25 (Conv2D) (None, 16, 16, 128) 147456 leaky_re_lu_19[0][0]
__________________________________________________________________________________________________
batch_normalization_20 (BatchNo (None, 16, 16, 128) 512 conv2d_25[0][0]
__________________________________________________________________________________________________
leaky_re_lu_20 (LeakyReLU) (None, 16, 16, 128) 0 batch_normalization_20[0][0]
__________________________________________________________________________________________________
conv2d_26 (Conv2D) (None, 16, 16, 128) 147456 leaky_re_lu_20[0][0]
__________________________________________________________________________________________________
add_5 (Add) (None, 16, 16, 128) 0 conv2d_26[0][0]
conv2d_23[0][0]
__________________________________________________________________________________________________
concatenate_5 (Concatenate) (None, 16, 16, 256) 0 add_5[0][0]
conv2d_26[0][0]
__________________________________________________________________________________________________
conv2d_27 (Conv2D) (None, 16, 16, 128) 32768 concatenate_5[0][0]
__________________________________________________________________________________________________
batch_normalization_21 (BatchNo (None, 16, 16, 128) 512 conv2d_27[0][0]
__________________________________________________________________________________________________
leaky_re_lu_21 (LeakyReLU) (None, 16, 16, 128) 0 batch_normalization_21[0][0]
__________________________________________________________________________________________________
conv2d_28 (Conv2D) (None, 16, 16, 1) 129 leaky_re_lu_21[0][0]
__________________________________________________________________________________________________
up_sampling2d (UpSampling2D) (None, 128, 128, 1) 0 conv2d_28[0][0]
==================================================================================================
Total params: 1,264,593
Trainable params: 1,261,425
Non-trainable params: 3,168
__________________________________________________________________________________________________
model summary: None
xxxxxxxxxx## We make a model with 28 Convolution layers with clubbed with leaky relu and batch normalisation layersxxxxxxxxxxfolder = PATH+'/stage_2_train_images'train_gen = generator(folder, train_filenames, pneumonia_locations, batch_size=16, image_size=128, shuffle=False, augment=True, predict=False)valid_gen = generator(folder, valid_filenames, pneumonia_locations, batch_size=16, image_size=128, shuffle=False, predict=False)history = model.fit_generator(train_gen, validation_data=valid_gen, callbacks=[learning_rate], epochs=10, shuffle=True, verbose=2)Epoch 1/10 WARNING:tensorflow:From /tensorflow-1.15.2/python3.7/tensorflow_core/python/ops/math_grad.py:1424: where (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version. Instructions for updating: 1507/1507 - 2646s - loss: 0.4735 - accuracy: 0.8636 - mean_iou: 0.6307 - val_loss: 0.6701 - val_accuracy: 0.6345 - val_mean_iou: 0.2398 - lr: 0.0010 Epoch 2/10 1507/1507 - 2277s - loss: 0.4450 - accuracy: 0.8668 - mean_iou: 0.6784 - val_loss: 0.4906 - val_accuracy: 0.6479 - val_mean_iou: 0.7674 - lr: 9.9606e-04 Epoch 3/10 1507/1507 - 2515s - loss: 0.4344 - accuracy: 0.9363 - mean_iou: 0.6933 - val_loss: 0.4465 - val_accuracy: 0.7194 - val_mean_iou: 0.7169 - lr: 9.8429e-04 Epoch 4/10 1507/1507 - 2618s - loss: 0.4286 - accuracy: 0.9385 - mean_iou: 0.6959 - val_loss: 0.4256 - val_accuracy: 0.7298 - val_mean_iou: 0.7254 - lr: 9.6489e-04 Epoch 5/10 1507/1507 - 2469s - loss: 0.4217 - accuracy: 0.9497 - mean_iou: 0.7069 - val_loss: 0.4358 - val_accuracy: 0.7421 - val_mean_iou: 0.7355 - lr: 9.3815e-04 Epoch 6/10 1507/1507 - 2573s - loss: 0.4173 - accuracy: 0.9400 - mean_iou: 0.7084 - val_loss: 0.4175 - val_accuracy: 0.7599 - val_mean_iou: 0.7172 - lr: 9.0451e-04 Epoch 7/10 1507/1507 - 2478s - loss: 0.4121 - accuracy: 0.9506 - mean_iou: 0.7106 - val_loss: 0.4145 - val_accuracy: 0.7731 - val_mean_iou: 0.7359 - lr: 8.6448e-04 Epoch 8/10 1507/1507 - 2290s - loss: 0.4099 - accuracy: 0.9609 - mean_iou: 0.7114 - val_loss: 0.4129 - val_accuracy: 0.7821 - val_mean_iou: 0.7054 - lr: 8.1871e-04 Epoch 9/10 1507/1507 - 2263s - loss: 0.4067 - accuracy: 0.9716 - mean_iou: 0.7151 - val_loss: 0.4147 - val_accuracy: 0.8019 - val_mean_iou: 0.7057 - lr: 7.6791e-04 Epoch 10/10 1507/1507 - 2262s - loss: 0.4044 - accuracy: 0.9716 - mean_iou: 0.7196 - val_loss: 0.4095 - val_accuracy: 0.8391 - val_mean_iou: 0.7293 - lr: 7.1289e-04
xxxxxxxxxxfor i in range(image_test_path): test_im_gray.append((np.dot(test_im[i], [0.2989, 0.5870, 0.1140])))test_im_gray = np.array(test_im_gray)test_labels = np.array(test_labels)test_loss, test_acc = model.evaluate(test_im_gray, test_labels, verbose=2)Test accuracy: 0.8124
xxxxxxxxxx## Next we train and test the model and see we have got Val accuracy of 0.8391 and test accuracy of 0.8124 whihch is not bad for scracth trained modelxxxxxxxxxxdf=pd.read_csv('/drive/MyDrive/GL_AIML/Capstone/stage_2_train_labels.csv')df['path']='/drive/MyDrive/GL_AIML/Capstone/stage_2_train_images/'+df['patientId'].astype(str)+'.dcm'xxxxxxxxxxtry: os.mkdir('/drive/MyDrive/GL_AIML/Capstone/data') os.mkdir('/drive/MyDrive/GL_AIML/Capstone/data/positive') os.mkdir('/drive/MyDrive/GL_AIML/Capstone/data/negative') os.chdir('/drive/MyDrive/GL_AIML/Capstone')except: passxxxxxxxxxxfor _,row in tqdm(unique_positive.iterrows()): img=pydicom.read_file(row['path']).pixel_array img=resize(img,(256,256)) plt.imsave('data/positive/'+row['patientId']+'.jpg',img,cmap='gray')for _,row in tqdm(negative.iterrows()): img=pydicom.read_file(row['path']).pixel_array img=resize(img,(256,256)) plt.imsave('data/negative/'+row['patientId']+'.jpg',img,cmap='gray')xxxxxxxxxxdatagen=ImageDataGenerator(samplewise_center=True,samplewise_std_normalization=True,horizontal_flip=True, width_shift_range=0.05,rescale=1/255,fill_mode='nearest',height_shift_range=0.05, preprocessing_function=preprocess_input,validation_split=0.3, )xxxxxxxxxxtrain=datagen.flow_from_directory('data',color_mode='rgb',batch_size=32, class_mode='binary',subset='training')test=datagen.flow_from_directory('data',color_mode='rgb',batch_size=32, class_mode='binary',subset='validation')xxxxxxxxxxpre_trained_model = VGG19(input_shape = (256,256,3), include_top = False, weights = 'imagenet')for layer in pre_trained_model.layers: layer.trainable = Falselast_layer = pre_trained_model.get_layer('block5_pool')print('last layer output shape: ', last_layer.output_shape)last_output = last_layer.outputDownloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg19/vgg19_weights_tf_dim_ordering_tf_kernels_notop.h5 80142336/80134624 [==============================] - 1s 0us/step last layer output shape: (None, 8, 8, 512)
xxxxxxxxxxmodel = Flatten()(last_output)model = Dense(1024)(model)model=LeakyReLU(0.1)(model)model=Dropout(0.25)(model)model=BatchNormalization()(model)model = Dense(1024)(model)model=LeakyReLU(0.1)(model)model=Dropout(0.25)(model)model=BatchNormalization()(model)model = Dense(1, activation='sigmoid')(model)xxxxxxxxxxfmodel = Model( pre_trained_model.input, model) fmodel.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])fmodel.summary()Model: "model" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= input_1 (InputLayer) [(None, 256, 256, 3)] 0 _________________________________________________________________ block1_conv1 (Conv2D) (None, 256, 256, 64) 1792 _________________________________________________________________ block1_conv2 (Conv2D) (None, 256, 256, 64) 36928 _________________________________________________________________ block1_pool (MaxPooling2D) (None, 128, 128, 64) 0 _________________________________________________________________ block2_conv1 (Conv2D) (None, 128, 128, 128) 73856 _________________________________________________________________ block2_conv2 (Conv2D) (None, 128, 128, 128) 147584 _________________________________________________________________ block2_pool (MaxPooling2D) (None, 64, 64, 128) 0 _________________________________________________________________ block3_conv1 (Conv2D) (None, 64, 64, 256) 295168 _________________________________________________________________ block3_conv2 (Conv2D) (None, 64, 64, 256) 590080 _________________________________________________________________ block3_conv3 (Conv2D) (None, 64, 64, 256) 590080 _________________________________________________________________ block3_conv4 (Conv2D) (None, 64, 64, 256) 590080 _________________________________________________________________ block3_pool (MaxPooling2D) (None, 32, 32, 256) 0 _________________________________________________________________ block4_conv1 (Conv2D) (None, 32, 32, 512) 1180160 _________________________________________________________________ block4_conv2 (Conv2D) (None, 32, 32, 512) 2359808 _________________________________________________________________ block4_conv3 (Conv2D) (None, 32, 32, 512) 2359808 _________________________________________________________________ block4_conv4 (Conv2D) (None, 32, 32, 512) 2359808 _________________________________________________________________ block4_pool (MaxPooling2D) (None, 16, 16, 512) 0 _________________________________________________________________ block5_conv1 (Conv2D) (None, 16, 16, 512) 2359808 _________________________________________________________________ block5_conv2 (Conv2D) (None, 16, 16, 512) 2359808 _________________________________________________________________ block5_conv3 (Conv2D) (None, 16, 16, 512) 2359808 _________________________________________________________________ block5_conv4 (Conv2D) (None, 16, 16, 512) 2359808 _________________________________________________________________ block5_pool (MaxPooling2D) (None, 8, 8, 512) 0 _________________________________________________________________ flatten (Flatten) (None, 32768) 0 _________________________________________________________________ dense (Dense) (None, 1024) 33555456 _________________________________________________________________ leaky_re_lu (LeakyReLU) (None, 1024) 0 _________________________________________________________________ dropout (Dropout) (None, 1024) 0 _________________________________________________________________ batch_normalization (BatchNo (None, 1024) 4096 _________________________________________________________________ dense_1 (Dense) (None, 1024) 1049600 _________________________________________________________________ leaky_re_lu_1 (LeakyReLU) (None, 1024) 0 _________________________________________________________________ dropout_1 (Dropout) (None, 1024) 0 _________________________________________________________________ batch_normalization_1 (Batch (None, 1024) 4096 _________________________________________________________________ dense_2 (Dense) (None, 1) 1025 ================================================================= Total params: 54,638,657 Trainable params: 34,610,177 Non-trainable params: 20,028,480 _________________________________________________________________
xxxxxxxxxxearly=EarlyStopping(monitor='accuracy',patience=3,mode='auto')reduce_lr = ReduceLROnPlateau(monitor='accuracy', factor=0.5, patience=2, verbose=1,cooldown=0, mode='auto',min_delta=0.0001, min_lr=1e-5)xxxxxxxxxxclass_weight={0:1,1:3.3}fmodel.fit(train,epochs=10,callbacks=[reduce_lr], steps_per_epoch=100,validation_data=test,class_weight=class_weight)Epoch 1/10 100/100 [==============================] - 7985s 162s/step - loss: 1.2630 - accuracy: 0.5153 - val_loss: 0.8171 - val_accuracy: 0.2253 Epoch 2/10 100/100 [==============================] - 7930s 160s/step - loss: 1.4530 - accuracy: 0.5392 - val_loss: 0.9142 - val_accuracy: 0.2366 Epoch 3/10 100/100 [==============================] - 7928s 163s/step - loss: 1.3823 - accuracy: 0.5643 - val_loss: 0.6749 - val_accuracy: 0.3747 Epoch 4/10 100/100 [==============================] - 7900s 160s/step - loss: 1.4660 - accuracy: 0.6153 - val_loss: 0.6178 - val_accuracy: 0.4153 Epoch 5/10 100/100 [==============================] - 7840s 159s/step - loss: 1.4550 - accuracy: 0.6891 - val_loss: 0.6744 - val_accuracy: 0.5561 Epoch 6/10 100/100 [==============================] - 7936s 161s/step - loss: 1.4023 - accuracy: 0.6949 - val_loss: 0.6045 - val_accuracy: 0.6087 Epoch 7/10 100/100 [==============================] - 7988s 160s/step - loss: 1.3342 - accuracy: 0.7183 - val_loss: 0.7171 - val_accuracy: 0.6553 Epoch 8/10 100/100 [==============================] - 7990s 161s/step - loss: 1.2510 - accuracy: 0.8812 - val_loss: 0.8172 - val_accuracy: 0.6643 Epoch 9/10 100/100 [==============================] - 7778s 161s/step - loss: 1.2823 - accuracy: 0.9043 - val_loss: 0.6849 - val_accuracy: 0.7047 Epoch 10/10 100/100 [==============================] - 7828s 156s/step - loss: 1.2243 - accuracy: 0.9125 - val_loss: 0.6089 - val_accuracy: 0.7447 Epoch 00003: ReduceLROnPlateau reducing learning rate to 0.0005000000237487257.
xxxxxxxxxxfor i in range(image_test_path): test_im_gray.append((np.dot(test_im[i], [0.2989, 0.5870, 0.1140])))test_im_gray = np.array(test_im_gray)test_labels = np.array(test_labels)test_loss, test_acc = fmodel.evaluate(test_im_gray, test_labels, verbose=2)Test accuracy: 0.7016
xxxxxxxxxxpre_trained_model = InceptionV3(input_shape = (256,256,3), include_top = False, weights = 'imagenet')for layer in pre_trained_model.layers: layer.trainable = Falsepre_trained_model.summary()Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/inception_v3/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5 87916544/87910968 [==============================] - 1s 0us/step Model: "inception_v3" __________________________________________________________________________________________________ Layer (type) Output Shape Param # Connected to ================================================================================================== input_2 (InputLayer) [(None, 256, 256, 3) 0 __________________________________________________________________________________________________ conv2d (Conv2D) (None, 127, 127, 32) 864 input_2[0][0] __________________________________________________________________________________________________ batch_normalization_2 (BatchNor (None, 127, 127, 32) 96 conv2d[0][0] __________________________________________________________________________________________________ activation (Activation) (None, 127, 127, 32) 0 batch_normalization_2[0][0] __________________________________________________________________________________________________ conv2d_1 (Conv2D) (None, 125, 125, 32) 9216 activation[0][0] __________________________________________________________________________________________________ batch_normalization_3 (BatchNor (None, 125, 125, 32) 96 conv2d_1[0][0] __________________________________________________________________________________________________ activation_1 (Activation) (None, 125, 125, 32) 0 batch_normalization_3[0][0] __________________________________________________________________________________________________ conv2d_2 (Conv2D) (None, 125, 125, 64) 18432 activation_1[0][0] __________________________________________________________________________________________________ batch_normalization_4 (BatchNor (None, 125, 125, 64) 192 conv2d_2[0][0] __________________________________________________________________________________________________ activation_2 (Activation) (None, 125, 125, 64) 0 batch_normalization_4[0][0] __________________________________________________________________________________________________ max_pooling2d (MaxPooling2D) (None, 62, 62, 64) 0 activation_2[0][0] __________________________________________________________________________________________________ conv2d_3 (Conv2D) (None, 62, 62, 80) 5120 max_pooling2d[0][0] __________________________________________________________________________________________________ batch_normalization_5 (BatchNor (None, 62, 62, 80) 240 conv2d_3[0][0] __________________________________________________________________________________________________ activation_3 (Activation) (None, 62, 62, 80) 0 batch_normalization_5[0][0] __________________________________________________________________________________________________ conv2d_4 (Conv2D) (None, 60, 60, 192) 138240 activation_3[0][0] __________________________________________________________________________________________________ batch_normalization_6 (BatchNor (None, 60, 60, 192) 576 conv2d_4[0][0] __________________________________________________________________________________________________ activation_4 (Activation) (None, 60, 60, 192) 0 batch_normalization_6[0][0] __________________________________________________________________________________________________ max_pooling2d_1 (MaxPooling2D) (None, 29, 29, 192) 0 activation_4[0][0] __________________________________________________________________________________________________ conv2d_8 (Conv2D) (None, 29, 29, 64) 12288 max_pooling2d_1[0][0] __________________________________________________________________________________________________ batch_normalization_10 (BatchNo (None, 29, 29, 64) 192 conv2d_8[0][0] __________________________________________________________________________________________________ activation_8 (Activation) (None, 29, 29, 64) 0 batch_normalization_10[0][0] __________________________________________________________________________________________________ conv2d_6 (Conv2D) (None, 29, 29, 48) 9216 max_pooling2d_1[0][0] __________________________________________________________________________________________________ conv2d_9 (Conv2D) (None, 29, 29, 96) 55296 activation_8[0][0] __________________________________________________________________________________________________ batch_normalization_8 (BatchNor (None, 29, 29, 48) 144 conv2d_6[0][0] __________________________________________________________________________________________________ batch_normalization_11 (BatchNo (None, 29, 29, 96) 288 conv2d_9[0][0] __________________________________________________________________________________________________ activation_6 (Activation) (None, 29, 29, 48) 0 batch_normalization_8[0][0] __________________________________________________________________________________________________ activation_9 (Activation) (None, 29, 29, 96) 0 batch_normalization_11[0][0] __________________________________________________________________________________________________ average_pooling2d (AveragePooli (None, 29, 29, 192) 0 max_pooling2d_1[0][0] __________________________________________________________________________________________________ conv2d_5 (Conv2D) (None, 29, 29, 64) 12288 max_pooling2d_1[0][0] __________________________________________________________________________________________________ conv2d_7 (Conv2D) (None, 29, 29, 64) 76800 activation_6[0][0] __________________________________________________________________________________________________ conv2d_10 (Conv2D) (None, 29, 29, 96) 82944 activation_9[0][0] __________________________________________________________________________________________________ conv2d_11 (Conv2D) (None, 29, 29, 32) 6144 average_pooling2d[0][0] __________________________________________________________________________________________________ batch_normalization_7 (BatchNor (None, 29, 29, 64) 192 conv2d_5[0][0] __________________________________________________________________________________________________ batch_normalization_9 (BatchNor (None, 29, 29, 64) 192 conv2d_7[0][0] __________________________________________________________________________________________________ batch_normalization_12 (BatchNo (None, 29, 29, 96) 288 conv2d_10[0][0] __________________________________________________________________________________________________ batch_normalization_13 (BatchNo (None, 29, 29, 32) 96 conv2d_11[0][0] __________________________________________________________________________________________________ activation_5 (Activation) (None, 29, 29, 64) 0 batch_normalization_7[0][0] __________________________________________________________________________________________________ activation_7 (Activation) (None, 29, 29, 64) 0 batch_normalization_9[0][0] __________________________________________________________________________________________________ activation_10 (Activation) (None, 29, 29, 96) 0 batch_normalization_12[0][0] __________________________________________________________________________________________________ activation_11 (Activation) (None, 29, 29, 32) 0 batch_normalization_13[0][0] __________________________________________________________________________________________________ mixed0 (Concatenate) (None, 29, 29, 256) 0 activation_5[0][0] activation_7[0][0] activation_10[0][0] activation_11[0][0] __________________________________________________________________________________________________ conv2d_15 (Conv2D) (None, 29, 29, 64) 16384 mixed0[0][0] __________________________________________________________________________________________________ batch_normalization_17 (BatchNo (None, 29, 29, 64) 192 conv2d_15[0][0] __________________________________________________________________________________________________ activation_15 (Activation) (None, 29, 29, 64) 0 batch_normalization_17[0][0] __________________________________________________________________________________________________ conv2d_13 (Conv2D) (None, 29, 29, 48) 12288 mixed0[0][0] __________________________________________________________________________________________________ conv2d_16 (Conv2D) (None, 29, 29, 96) 55296 activation_15[0][0] __________________________________________________________________________________________________ batch_normalization_15 (BatchNo (None, 29, 29, 48) 144 conv2d_13[0][0] __________________________________________________________________________________________________ batch_normalization_18 (BatchNo (None, 29, 29, 96) 288 conv2d_16[0][0] __________________________________________________________________________________________________ activation_13 (Activation) (None, 29, 29, 48) 0 batch_normalization_15[0][0] __________________________________________________________________________________________________ activation_16 (Activation) (None, 29, 29, 96) 0 batch_normalization_18[0][0] __________________________________________________________________________________________________ average_pooling2d_1 (AveragePoo (None, 29, 29, 256) 0 mixed0[0][0] __________________________________________________________________________________________________ conv2d_12 (Conv2D) (None, 29, 29, 64) 16384 mixed0[0][0] __________________________________________________________________________________________________ conv2d_14 (Conv2D) (None, 29, 29, 64) 76800 activation_13[0][0] __________________________________________________________________________________________________ conv2d_17 (Conv2D) (None, 29, 29, 96) 82944 activation_16[0][0] __________________________________________________________________________________________________ conv2d_18 (Conv2D) (None, 29, 29, 64) 16384 average_pooling2d_1[0][0] __________________________________________________________________________________________________ batch_normalization_14 (BatchNo (None, 29, 29, 64) 192 conv2d_12[0][0] __________________________________________________________________________________________________ batch_normalization_16 (BatchNo (None, 29, 29, 64) 192 conv2d_14[0][0] __________________________________________________________________________________________________ batch_normalization_19 (BatchNo (None, 29, 29, 96) 288 conv2d_17[0][0] __________________________________________________________________________________________________ batch_normalization_20 (BatchNo (None, 29, 29, 64) 192 conv2d_18[0][0] __________________________________________________________________________________________________ activation_12 (Activation) (None, 29, 29, 64) 0 batch_normalization_14[0][0] __________________________________________________________________________________________________ activation_14 (Activation) (None, 29, 29, 64) 0 batch_normalization_16[0][0] __________________________________________________________________________________________________ activation_17 (Activation) (None, 29, 29, 96) 0 batch_normalization_19[0][0] __________________________________________________________________________________________________ activation_18 (Activation) (None, 29, 29, 64) 0 batch_normalization_20[0][0] __________________________________________________________________________________________________ mixed1 (Concatenate) (None, 29, 29, 288) 0 activation_12[0][0] activation_14[0][0] activation_17[0][0] activation_18[0][0] __________________________________________________________________________________________________ conv2d_22 (Conv2D) (None, 29, 29, 64) 18432 mixed1[0][0] __________________________________________________________________________________________________ batch_normalization_24 (BatchNo (None, 29, 29, 64) 192 conv2d_22[0][0] __________________________________________________________________________________________________ activation_22 (Activation) (None, 29, 29, 64) 0 batch_normalization_24[0][0] __________________________________________________________________________________________________ conv2d_20 (Conv2D) (None, 29, 29, 48) 13824 mixed1[0][0] __________________________________________________________________________________________________ conv2d_23 (Conv2D) (None, 29, 29, 96) 55296 activation_22[0][0] __________________________________________________________________________________________________ batch_normalization_22 (BatchNo (None, 29, 29, 48) 144 conv2d_20[0][0] __________________________________________________________________________________________________ batch_normalization_25 (BatchNo (None, 29, 29, 96) 288 conv2d_23[0][0] __________________________________________________________________________________________________ activation_20 (Activation) (None, 29, 29, 48) 0 batch_normalization_22[0][0] __________________________________________________________________________________________________ activation_23 (Activation) (None, 29, 29, 96) 0 batch_normalization_25[0][0] __________________________________________________________________________________________________ average_pooling2d_2 (AveragePoo (None, 29, 29, 288) 0 mixed1[0][0] __________________________________________________________________________________________________ conv2d_19 (Conv2D) (None, 29, 29, 64) 18432 mixed1[0][0] __________________________________________________________________________________________________ conv2d_21 (Conv2D) (None, 29, 29, 64) 76800 activation_20[0][0] __________________________________________________________________________________________________ conv2d_24 (Conv2D) (None, 29, 29, 96) 82944 activation_23[0][0] __________________________________________________________________________________________________ conv2d_25 (Conv2D) (None, 29, 29, 64) 18432 average_pooling2d_2[0][0] __________________________________________________________________________________________________ batch_normalization_21 (BatchNo (None, 29, 29, 64) 192 conv2d_19[0][0] __________________________________________________________________________________________________ batch_normalization_23 (BatchNo (None, 29, 29, 64) 192 conv2d_21[0][0] __________________________________________________________________________________________________ batch_normalization_26 (BatchNo (None, 29, 29, 96) 288 conv2d_24[0][0] __________________________________________________________________________________________________ batch_normalization_27 (BatchNo (None, 29, 29, 64) 192 conv2d_25[0][0] __________________________________________________________________________________________________ activation_19 (Activation) (None, 29, 29, 64) 0 batch_normalization_21[0][0] __________________________________________________________________________________________________ activation_21 (Activation) (None, 29, 29, 64) 0 batch_normalization_23[0][0] __________________________________________________________________________________________________ activation_24 (Activation) (None, 29, 29, 96) 0 batch_normalization_26[0][0] __________________________________________________________________________________________________ activation_25 (Activation) (None, 29, 29, 64) 0 batch_normalization_27[0][0] __________________________________________________________________________________________________ mixed2 (Concatenate) (None, 29, 29, 288) 0 activation_19[0][0] activation_21[0][0] activation_24[0][0] activation_25[0][0] __________________________________________________________________________________________________ conv2d_27 (Conv2D) (None, 29, 29, 64) 18432 mixed2[0][0] __________________________________________________________________________________________________ batch_normalization_29 (BatchNo (None, 29, 29, 64) 192 conv2d_27[0][0] __________________________________________________________________________________________________ activation_27 (Activation) (None, 29, 29, 64) 0 batch_normalization_29[0][0] __________________________________________________________________________________________________ conv2d_28 (Conv2D) (None, 29, 29, 96) 55296 activation_27[0][0] __________________________________________________________________________________________________ batch_normalization_30 (BatchNo (None, 29, 29, 96) 288 conv2d_28[0][0] __________________________________________________________________________________________________ activation_28 (Activation) (None, 29, 29, 96) 0 batch_normalization_30[0][0] __________________________________________________________________________________________________ conv2d_26 (Conv2D) (None, 14, 14, 384) 995328 mixed2[0][0] __________________________________________________________________________________________________ conv2d_29 (Conv2D) (None, 14, 14, 96) 82944 activation_28[0][0] __________________________________________________________________________________________________ batch_normalization_28 (BatchNo (None, 14, 14, 384) 1152 conv2d_26[0][0] __________________________________________________________________________________________________ batch_normalization_31 (BatchNo (None, 14, 14, 96) 288 conv2d_29[0][0] __________________________________________________________________________________________________ activation_26 (Activation) (None, 14, 14, 384) 0 batch_normalization_28[0][0] __________________________________________________________________________________________________ activation_29 (Activation) (None, 14, 14, 96) 0 batch_normalization_31[0][0] __________________________________________________________________________________________________ max_pooling2d_2 (MaxPooling2D) (None, 14, 14, 288) 0 mixed2[0][0] __________________________________________________________________________________________________ mixed3 (Concatenate) (None, 14, 14, 768) 0 activation_26[0][0] activation_29[0][0] max_pooling2d_2[0][0] __________________________________________________________________________________________________ conv2d_34 (Conv2D) (None, 14, 14, 128) 98304 mixed3[0][0] __________________________________________________________________________________________________ batch_normalization_36 (BatchNo (None, 14, 14, 128) 384 conv2d_34[0][0] __________________________________________________________________________________________________ activation_34 (Activation) (None, 14, 14, 128) 0 batch_normalization_36[0][0] __________________________________________________________________________________________________ conv2d_35 (Conv2D) (None, 14, 14, 128) 114688 activation_34[0][0] __________________________________________________________________________________________________ batch_normalization_37 (BatchNo (None, 14, 14, 128) 384 conv2d_35[0][0] __________________________________________________________________________________________________ activation_35 (Activation) (None, 14, 14, 128) 0 batch_normalization_37[0][0] __________________________________________________________________________________________________ conv2d_31 (Conv2D) (None, 14, 14, 128) 98304 mixed3[0][0] __________________________________________________________________________________________________ conv2d_36 (Conv2D) (None, 14, 14, 128) 114688 activation_35[0][0] __________________________________________________________________________________________________ batch_normalization_33 (BatchNo (None, 14, 14, 128) 384 conv2d_31[0][0] __________________________________________________________________________________________________ batch_normalization_38 (BatchNo (None, 14, 14, 128) 384 conv2d_36[0][0] __________________________________________________________________________________________________ activation_31 (Activation) (None, 14, 14, 128) 0 batch_normalization_33[0][0] __________________________________________________________________________________________________ activation_36 (Activation) (None, 14, 14, 128) 0 batch_normalization_38[0][0] __________________________________________________________________________________________________ conv2d_32 (Conv2D) (None, 14, 14, 128) 114688 activation_31[0][0] __________________________________________________________________________________________________ conv2d_37 (Conv2D) (None, 14, 14, 128) 114688 activation_36[0][0] __________________________________________________________________________________________________ batch_normalization_34 (BatchNo (None, 14, 14, 128) 384 conv2d_32[0][0] __________________________________________________________________________________________________ batch_normalization_39 (BatchNo (None, 14, 14, 128) 384 conv2d_37[0][0] __________________________________________________________________________________________________ activation_32 (Activation) (None, 14, 14, 128) 0 batch_normalization_34[0][0] __________________________________________________________________________________________________ activation_37 (Activation) (None, 14, 14, 128) 0 batch_normalization_39[0][0] __________________________________________________________________________________________________ average_pooling2d_3 (AveragePoo (None, 14, 14, 768) 0 mixed3[0][0] __________________________________________________________________________________________________ conv2d_30 (Conv2D) (None, 14, 14, 192) 147456 mixed3[0][0] __________________________________________________________________________________________________ conv2d_33 (Conv2D) (None, 14, 14, 192) 172032 activation_32[0][0] __________________________________________________________________________________________________ conv2d_38 (Conv2D) (None, 14, 14, 192) 172032 activation_37[0][0] __________________________________________________________________________________________________ conv2d_39 (Conv2D) (None, 14, 14, 192) 147456 average_pooling2d_3[0][0] __________________________________________________________________________________________________ batch_normalization_32 (BatchNo (None, 14, 14, 192) 576 conv2d_30[0][0] __________________________________________________________________________________________________ batch_normalization_35 (BatchNo (None, 14, 14, 192) 576 conv2d_33[0][0] __________________________________________________________________________________________________ batch_normalization_40 (BatchNo (None, 14, 14, 192) 576 conv2d_38[0][0] __________________________________________________________________________________________________ batch_normalization_41 (BatchNo (None, 14, 14, 192) 576 conv2d_39[0][0] __________________________________________________________________________________________________ activation_30 (Activation) (None, 14, 14, 192) 0 batch_normalization_32[0][0] __________________________________________________________________________________________________ activation_33 (Activation) (None, 14, 14, 192) 0 batch_normalization_35[0][0] __________________________________________________________________________________________________ activation_38 (Activation) (None, 14, 14, 192) 0 batch_normalization_40[0][0] __________________________________________________________________________________________________ activation_39 (Activation) (None, 14, 14, 192) 0 batch_normalization_41[0][0] __________________________________________________________________________________________________ mixed4 (Concatenate) (None, 14, 14, 768) 0 activation_30[0][0] activation_33[0][0] activation_38[0][0] activation_39[0][0] __________________________________________________________________________________________________ conv2d_44 (Conv2D) (None, 14, 14, 160) 122880 mixed4[0][0] __________________________________________________________________________________________________ batch_normalization_46 (BatchNo (None, 14, 14, 160) 480 conv2d_44[0][0] __________________________________________________________________________________________________ activation_44 (Activation) (None, 14, 14, 160) 0 batch_normalization_46[0][0] __________________________________________________________________________________________________ conv2d_45 (Conv2D) (None, 14, 14, 160) 179200 activation_44[0][0] __________________________________________________________________________________________________ batch_normalization_47 (BatchNo (None, 14, 14, 160) 480 conv2d_45[0][0] __________________________________________________________________________________________________ activation_45 (Activation) (None, 14, 14, 160) 0 batch_normalization_47[0][0] __________________________________________________________________________________________________ conv2d_41 (Conv2D) (None, 14, 14, 160) 122880 mixed4[0][0] __________________________________________________________________________________________________ conv2d_46 (Conv2D) (None, 14, 14, 160) 179200 activation_45[0][0] __________________________________________________________________________________________________ batch_normalization_43 (BatchNo (None, 14, 14, 160) 480 conv2d_41[0][0] __________________________________________________________________________________________________ batch_normalization_48 (BatchNo (None, 14, 14, 160) 480 conv2d_46[0][0] __________________________________________________________________________________________________ activation_41 (Activation) (None, 14, 14, 160) 0 batch_normalization_43[0][0] __________________________________________________________________________________________________ activation_46 (Activation) (None, 14, 14, 160) 0 batch_normalization_48[0][0] __________________________________________________________________________________________________ conv2d_42 (Conv2D) (None, 14, 14, 160) 179200 activation_41[0][0] __________________________________________________________________________________________________ conv2d_47 (Conv2D) (None, 14, 14, 160) 179200 activation_46[0][0] __________________________________________________________________________________________________ batch_normalization_44 (BatchNo (None, 14, 14, 160) 480 conv2d_42[0][0] __________________________________________________________________________________________________ batch_normalization_49 (BatchNo (None, 14, 14, 160) 480 conv2d_47[0][0] __________________________________________________________________________________________________ activation_42 (Activation) (None, 14, 14, 160) 0 batch_normalization_44[0][0] __________________________________________________________________________________________________ activation_47 (Activation) (None, 14, 14, 160) 0 batch_normalization_49[0][0] __________________________________________________________________________________________________ average_pooling2d_4 (AveragePoo (None, 14, 14, 768) 0 mixed4[0][0] __________________________________________________________________________________________________ conv2d_40 (Conv2D) (None, 14, 14, 192) 147456 mixed4[0][0] __________________________________________________________________________________________________ conv2d_43 (Conv2D) (None, 14, 14, 192) 215040 activation_42[0][0] __________________________________________________________________________________________________ conv2d_48 (Conv2D) (None, 14, 14, 192) 215040 activation_47[0][0] __________________________________________________________________________________________________ conv2d_49 (Conv2D) (None, 14, 14, 192) 147456 average_pooling2d_4[0][0] __________________________________________________________________________________________________ batch_normalization_42 (BatchNo (None, 14, 14, 192) 576 conv2d_40[0][0] __________________________________________________________________________________________________ batch_normalization_45 (BatchNo (None, 14, 14, 192) 576 conv2d_43[0][0] __________________________________________________________________________________________________ batch_normalization_50 (BatchNo (None, 14, 14, 192) 576 conv2d_48[0][0] __________________________________________________________________________________________________ batch_normalization_51 (BatchNo (None, 14, 14, 192) 576 conv2d_49[0][0] __________________________________________________________________________________________________ activation_40 (Activation) (None, 14, 14, 192) 0 batch_normalization_42[0][0] __________________________________________________________________________________________________ activation_43 (Activation) (None, 14, 14, 192) 0 batch_normalization_45[0][0] __________________________________________________________________________________________________ activation_48 (Activation) (None, 14, 14, 192) 0 batch_normalization_50[0][0] __________________________________________________________________________________________________ activation_49 (Activation) (None, 14, 14, 192) 0 batch_normalization_51[0][0] __________________________________________________________________________________________________ mixed5 (Concatenate) (None, 14, 14, 768) 0 activation_40[0][0] activation_43[0][0] activation_48[0][0] activation_49[0][0] __________________________________________________________________________________________________ conv2d_54 (Conv2D) (None, 14, 14, 160) 122880 mixed5[0][0] __________________________________________________________________________________________________ batch_normalization_56 (BatchNo (None, 14, 14, 160) 480 conv2d_54[0][0] __________________________________________________________________________________________________ activation_54 (Activation) (None, 14, 14, 160) 0 batch_normalization_56[0][0] __________________________________________________________________________________________________ conv2d_55 (Conv2D) (None, 14, 14, 160) 179200 activation_54[0][0] __________________________________________________________________________________________________ batch_normalization_57 (BatchNo (None, 14, 14, 160) 480 conv2d_55[0][0] __________________________________________________________________________________________________ activation_55 (Activation) (None, 14, 14, 160) 0 batch_normalization_57[0][0] __________________________________________________________________________________________________ conv2d_51 (Conv2D) (None, 14, 14, 160) 122880 mixed5[0][0] __________________________________________________________________________________________________ conv2d_56 (Conv2D) (None, 14, 14, 160) 179200 activation_55[0][0] __________________________________________________________________________________________________ batch_normalization_53 (BatchNo (None, 14, 14, 160) 480 conv2d_51[0][0] __________________________________________________________________________________________________ batch_normalization_58 (BatchNo (None, 14, 14, 160) 480 conv2d_56[0][0] __________________________________________________________________________________________________ activation_51 (Activation) (None, 14, 14, 160) 0 batch_normalization_53[0][0] __________________________________________________________________________________________________ activation_56 (Activation) (None, 14, 14, 160) 0 batch_normalization_58[0][0] __________________________________________________________________________________________________ conv2d_52 (Conv2D) (None, 14, 14, 160) 179200 activation_51[0][0] __________________________________________________________________________________________________ conv2d_57 (Conv2D) (None, 14, 14, 160) 179200 activation_56[0][0] __________________________________________________________________________________________________ batch_normalization_54 (BatchNo (None, 14, 14, 160) 480 conv2d_52[0][0] __________________________________________________________________________________________________ batch_normalization_59 (BatchNo (None, 14, 14, 160) 480 conv2d_57[0][0] __________________________________________________________________________________________________ activation_52 (Activation) (None, 14, 14, 160) 0 batch_normalization_54[0][0] __________________________________________________________________________________________________ activation_57 (Activation) (None, 14, 14, 160) 0 batch_normalization_59[0][0] __________________________________________________________________________________________________ average_pooling2d_5 (AveragePoo (None, 14, 14, 768) 0 mixed5[0][0] __________________________________________________________________________________________________ conv2d_50 (Conv2D) (None, 14, 14, 192) 147456 mixed5[0][0] __________________________________________________________________________________________________ conv2d_53 (Conv2D) (None, 14, 14, 192) 215040 activation_52[0][0] __________________________________________________________________________________________________ conv2d_58 (Conv2D) (None, 14, 14, 192) 215040 activation_57[0][0] __________________________________________________________________________________________________ conv2d_59 (Conv2D) (None, 14, 14, 192) 147456 average_pooling2d_5[0][0] __________________________________________________________________________________________________ batch_normalization_52 (BatchNo (None, 14, 14, 192) 576 conv2d_50[0][0] __________________________________________________________________________________________________ batch_normalization_55 (BatchNo (None, 14, 14, 192) 576 conv2d_53[0][0] __________________________________________________________________________________________________ batch_normalization_60 (BatchNo (None, 14, 14, 192) 576 conv2d_58[0][0] __________________________________________________________________________________________________ batch_normalization_61 (BatchNo (None, 14, 14, 192) 576 conv2d_59[0][0] __________________________________________________________________________________________________ activation_50 (Activation) (None, 14, 14, 192) 0 batch_normalization_52[0][0] __________________________________________________________________________________________________ activation_53 (Activation) (None, 14, 14, 192) 0 batch_normalization_55[0][0] __________________________________________________________________________________________________ activation_58 (Activation) (None, 14, 14, 192) 0 batch_normalization_60[0][0] __________________________________________________________________________________________________ activation_59 (Activation) (None, 14, 14, 192) 0 batch_normalization_61[0][0] __________________________________________________________________________________________________ mixed6 (Concatenate) (None, 14, 14, 768) 0 activation_50[0][0] activation_53[0][0] activation_58[0][0] activation_59[0][0] __________________________________________________________________________________________________ conv2d_64 (Conv2D) (None, 14, 14, 192) 147456 mixed6[0][0] __________________________________________________________________________________________________ batch_normalization_66 (BatchNo (None, 14, 14, 192) 576 conv2d_64[0][0] __________________________________________________________________________________________________ activation_64 (Activation) (None, 14, 14, 192) 0 batch_normalization_66[0][0] __________________________________________________________________________________________________ conv2d_65 (Conv2D) (None, 14, 14, 192) 258048 activation_64[0][0] __________________________________________________________________________________________________ batch_normalization_67 (BatchNo (None, 14, 14, 192) 576 conv2d_65[0][0] __________________________________________________________________________________________________ activation_65 (Activation) (None, 14, 14, 192) 0 batch_normalization_67[0][0] __________________________________________________________________________________________________ conv2d_61 (Conv2D) (None, 14, 14, 192) 147456 mixed6[0][0] __________________________________________________________________________________________________ conv2d_66 (Conv2D) (None, 14, 14, 192) 258048 activation_65[0][0] __________________________________________________________________________________________________ batch_normalization_63 (BatchNo (None, 14, 14, 192) 576 conv2d_61[0][0] __________________________________________________________________________________________________ batch_normalization_68 (BatchNo (None, 14, 14, 192) 576 conv2d_66[0][0] __________________________________________________________________________________________________ activation_61 (Activation) (None, 14, 14, 192) 0 batch_normalization_63[0][0] __________________________________________________________________________________________________ activation_66 (Activation) (None, 14, 14, 192) 0 batch_normalization_68[0][0] __________________________________________________________________________________________________ conv2d_62 (Conv2D) (None, 14, 14, 192) 258048 activation_61[0][0] __________________________________________________________________________________________________ conv2d_67 (Conv2D) (None, 14, 14, 192) 258048 activation_66[0][0] __________________________________________________________________________________________________ batch_normalization_64 (BatchNo (None, 14, 14, 192) 576 conv2d_62[0][0] __________________________________________________________________________________________________ batch_normalization_69 (BatchNo (None, 14, 14, 192) 576 conv2d_67[0][0] __________________________________________________________________________________________________ activation_62 (Activation) (None, 14, 14, 192) 0 batch_normalization_64[0][0] __________________________________________________________________________________________________ activation_67 (Activation) (None, 14, 14, 192) 0 batch_normalization_69[0][0] __________________________________________________________________________________________________ average_pooling2d_6 (AveragePoo (None, 14, 14, 768) 0 mixed6[0][0] __________________________________________________________________________________________________ conv2d_60 (Conv2D) (None, 14, 14, 192) 147456 mixed6[0][0] __________________________________________________________________________________________________ conv2d_63 (Conv2D) (None, 14, 14, 192) 258048 activation_62[0][0] __________________________________________________________________________________________________ conv2d_68 (Conv2D) (None, 14, 14, 192) 258048 activation_67[0][0] __________________________________________________________________________________________________ conv2d_69 (Conv2D) (None, 14, 14, 192) 147456 average_pooling2d_6[0][0] __________________________________________________________________________________________________ batch_normalization_62 (BatchNo (None, 14, 14, 192) 576 conv2d_60[0][0] __________________________________________________________________________________________________ batch_normalization_65 (BatchNo (None, 14, 14, 192) 576 conv2d_63[0][0] __________________________________________________________________________________________________ batch_normalization_70 (BatchNo (None, 14, 14, 192) 576 conv2d_68[0][0] __________________________________________________________________________________________________ batch_normalization_71 (BatchNo (None, 14, 14, 192) 576 conv2d_69[0][0] __________________________________________________________________________________________________ activation_60 (Activation) (None, 14, 14, 192) 0 batch_normalization_62[0][0] __________________________________________________________________________________________________ activation_63 (Activation) (None, 14, 14, 192) 0 batch_normalization_65[0][0] __________________________________________________________________________________________________ activation_68 (Activation) (None, 14, 14, 192) 0 batch_normalization_70[0][0] __________________________________________________________________________________________________ activation_69 (Activation) (None, 14, 14, 192) 0 batch_normalization_71[0][0] __________________________________________________________________________________________________ mixed7 (Concatenate) (None, 14, 14, 768) 0 activation_60[0][0] activation_63[0][0] activation_68[0][0] activation_69[0][0] __________________________________________________________________________________________________ conv2d_72 (Conv2D) (None, 14, 14, 192) 147456 mixed7[0][0] __________________________________________________________________________________________________ batch_normalization_74 (BatchNo (None, 14, 14, 192) 576 conv2d_72[0][0] __________________________________________________________________________________________________ activation_72 (Activation) (None, 14, 14, 192) 0 batch_normalization_74[0][0] __________________________________________________________________________________________________ conv2d_73 (Conv2D) (None, 14, 14, 192) 258048 activation_72[0][0] __________________________________________________________________________________________________ batch_normalization_75 (BatchNo (None, 14, 14, 192) 576 conv2d_73[0][0] __________________________________________________________________________________________________ activation_73 (Activation) (None, 14, 14, 192) 0 batch_normalization_75[0][0] __________________________________________________________________________________________________ conv2d_70 (Conv2D) (None, 14, 14, 192) 147456 mixed7[0][0] __________________________________________________________________________________________________ conv2d_74 (Conv2D) (None, 14, 14, 192) 258048 activation_73[0][0] __________________________________________________________________________________________________ batch_normalization_72 (BatchNo (None, 14, 14, 192) 576 conv2d_70[0][0] __________________________________________________________________________________________________ batch_normalization_76 (BatchNo (None, 14, 14, 192) 576 conv2d_74[0][0] __________________________________________________________________________________________________ activation_70 (Activation) (None, 14, 14, 192) 0 batch_normalization_72[0][0] __________________________________________________________________________________________________ activation_74 (Activation) (None, 14, 14, 192) 0 batch_normalization_76[0][0] __________________________________________________________________________________________________ conv2d_71 (Conv2D) (None, 6, 6, 320) 552960 activation_70[0][0] __________________________________________________________________________________________________ conv2d_75 (Conv2D) (None, 6, 6, 192) 331776 activation_74[0][0] __________________________________________________________________________________________________ batch_normalization_73 (BatchNo (None, 6, 6, 320) 960 conv2d_71[0][0] __________________________________________________________________________________________________ batch_normalization_77 (BatchNo (None, 6, 6, 192) 576 conv2d_75[0][0] __________________________________________________________________________________________________ activation_71 (Activation) (None, 6, 6, 320) 0 batch_normalization_73[0][0] __________________________________________________________________________________________________ activation_75 (Activation) (None, 6, 6, 192) 0 batch_normalization_77[0][0] __________________________________________________________________________________________________ max_pooling2d_3 (MaxPooling2D) (None, 6, 6, 768) 0 mixed7[0][0] __________________________________________________________________________________________________ mixed8 (Concatenate) (None, 6, 6, 1280) 0 activation_71[0][0] activation_75[0][0] max_pooling2d_3[0][0] __________________________________________________________________________________________________ conv2d_80 (Conv2D) (None, 6, 6, 448) 573440 mixed8[0][0] __________________________________________________________________________________________________ batch_normalization_82 (BatchNo (None, 6, 6, 448) 1344 conv2d_80[0][0] __________________________________________________________________________________________________ activation_80 (Activation) (None, 6, 6, 448) 0 batch_normalization_82[0][0] __________________________________________________________________________________________________ conv2d_77 (Conv2D) (None, 6, 6, 384) 491520 mixed8[0][0] __________________________________________________________________________________________________ conv2d_81 (Conv2D) (None, 6, 6, 384) 1548288 activation_80[0][0] __________________________________________________________________________________________________ batch_normalization_79 (BatchNo (None, 6, 6, 384) 1152 conv2d_77[0][0] __________________________________________________________________________________________________ batch_normalization_83 (BatchNo (None, 6, 6, 384) 1152 conv2d_81[0][0] __________________________________________________________________________________________________ activation_77 (Activation) (None, 6, 6, 384) 0 batch_normalization_79[0][0] __________________________________________________________________________________________________ activation_81 (Activation) (None, 6, 6, 384) 0 batch_normalization_83[0][0] __________________________________________________________________________________________________ conv2d_78 (Conv2D) (None, 6, 6, 384) 442368 activation_77[0][0] __________________________________________________________________________________________________ conv2d_79 (Conv2D) (None, 6, 6, 384) 442368 activation_77[0][0] __________________________________________________________________________________________________ conv2d_82 (Conv2D) (None, 6, 6, 384) 442368 activation_81[0][0] __________________________________________________________________________________________________ conv2d_83 (Conv2D) (None, 6, 6, 384) 442368 activation_81[0][0] __________________________________________________________________________________________________ average_pooling2d_7 (AveragePoo (None, 6, 6, 1280) 0 mixed8[0][0] __________________________________________________________________________________________________ conv2d_76 (Conv2D) (None, 6, 6, 320) 409600 mixed8[0][0] __________________________________________________________________________________________________ batch_normalization_80 (BatchNo (None, 6, 6, 384) 1152 conv2d_78[0][0] __________________________________________________________________________________________________ batch_normalization_81 (BatchNo (None, 6, 6, 384) 1152 conv2d_79[0][0] __________________________________________________________________________________________________ batch_normalization_84 (BatchNo (None, 6, 6, 384) 1152 conv2d_82[0][0] __________________________________________________________________________________________________ batch_normalization_85 (BatchNo (None, 6, 6, 384) 1152 conv2d_83[0][0] __________________________________________________________________________________________________ conv2d_84 (Conv2D) (None, 6, 6, 192) 245760 average_pooling2d_7[0][0] __________________________________________________________________________________________________ batch_normalization_78 (BatchNo (None, 6, 6, 320) 960 conv2d_76[0][0] __________________________________________________________________________________________________ activation_78 (Activation) (None, 6, 6, 384) 0 batch_normalization_80[0][0] __________________________________________________________________________________________________ activation_79 (Activation) (None, 6, 6, 384) 0 batch_normalization_81[0][0] __________________________________________________________________________________________________ activation_82 (Activation) (None, 6, 6, 384) 0 batch_normalization_84[0][0] __________________________________________________________________________________________________ activation_83 (Activation) (None, 6, 6, 384) 0 batch_normalization_85[0][0] __________________________________________________________________________________________________ batch_normalization_86 (BatchNo (None, 6, 6, 192) 576 conv2d_84[0][0] __________________________________________________________________________________________________ activation_76 (Activation) (None, 6, 6, 320) 0 batch_normalization_78[0][0] __________________________________________________________________________________________________ mixed9_0 (Concatenate) (None, 6, 6, 768) 0 activation_78[0][0] activation_79[0][0] __________________________________________________________________________________________________ concatenate (Concatenate) (None, 6, 6, 768) 0 activation_82[0][0] activation_83[0][0] __________________________________________________________________________________________________ activation_84 (Activation) (None, 6, 6, 192) 0 batch_normalization_86[0][0] __________________________________________________________________________________________________ mixed9 (Concatenate) (None, 6, 6, 2048) 0 activation_76[0][0] mixed9_0[0][0] concatenate[0][0] activation_84[0][0] __________________________________________________________________________________________________ conv2d_89 (Conv2D) (None, 6, 6, 448) 917504 mixed9[0][0] __________________________________________________________________________________________________ batch_normalization_91 (BatchNo (None, 6, 6, 448) 1344 conv2d_89[0][0] __________________________________________________________________________________________________ activation_89 (Activation) (None, 6, 6, 448) 0 batch_normalization_91[0][0] __________________________________________________________________________________________________ conv2d_86 (Conv2D) (None, 6, 6, 384) 786432 mixed9[0][0] __________________________________________________________________________________________________ conv2d_90 (Conv2D) (None, 6, 6, 384) 1548288 activation_89[0][0] __________________________________________________________________________________________________ batch_normalization_88 (BatchNo (None, 6, 6, 384) 1152 conv2d_86[0][0] __________________________________________________________________________________________________ batch_normalization_92 (BatchNo (None, 6, 6, 384) 1152 conv2d_90[0][0] __________________________________________________________________________________________________ activation_86 (Activation) (None, 6, 6, 384) 0 batch_normalization_88[0][0] __________________________________________________________________________________________________ activation_90 (Activation) (None, 6, 6, 384) 0 batch_normalization_92[0][0] __________________________________________________________________________________________________ conv2d_87 (Conv2D) (None, 6, 6, 384) 442368 activation_86[0][0] __________________________________________________________________________________________________ conv2d_88 (Conv2D) (None, 6, 6, 384) 442368 activation_86[0][0] __________________________________________________________________________________________________ conv2d_91 (Conv2D) (None, 6, 6, 384) 442368 activation_90[0][0] __________________________________________________________________________________________________ conv2d_92 (Conv2D) (None, 6, 6, 384) 442368 activation_90[0][0] __________________________________________________________________________________________________ average_pooling2d_8 (AveragePoo (None, 6, 6, 2048) 0 mixed9[0][0] __________________________________________________________________________________________________ conv2d_85 (Conv2D) (None, 6, 6, 320) 655360 mixed9[0][0] __________________________________________________________________________________________________ batch_normalization_89 (BatchNo (None, 6, 6, 384) 1152 conv2d_87[0][0] __________________________________________________________________________________________________ batch_normalization_90 (BatchNo (None, 6, 6, 384) 1152 conv2d_88[0][0] __________________________________________________________________________________________________ batch_normalization_93 (BatchNo (None, 6, 6, 384) 1152 conv2d_91[0][0] __________________________________________________________________________________________________ batch_normalization_94 (BatchNo (None, 6, 6, 384) 1152 conv2d_92[0][0] __________________________________________________________________________________________________ conv2d_93 (Conv2D) (None, 6, 6, 192) 393216 average_pooling2d_8[0][0] __________________________________________________________________________________________________ batch_normalization_87 (BatchNo (None, 6, 6, 320) 960 conv2d_85[0][0] __________________________________________________________________________________________________ activation_87 (Activation) (None, 6, 6, 384) 0 batch_normalization_89[0][0] __________________________________________________________________________________________________ activation_88 (Activation) (None, 6, 6, 384) 0 batch_normalization_90[0][0] __________________________________________________________________________________________________ activation_91 (Activation) (None, 6, 6, 384) 0 batch_normalization_93[0][0] __________________________________________________________________________________________________ activation_92 (Activation) (None, 6, 6, 384) 0 batch_normalization_94[0][0] __________________________________________________________________________________________________ batch_normalization_95 (BatchNo (None, 6, 6, 192) 576 conv2d_93[0][0] __________________________________________________________________________________________________ activation_85 (Activation) (None, 6, 6, 320) 0 batch_normalization_87[0][0] __________________________________________________________________________________________________ mixed9_1 (Concatenate) (None, 6, 6, 768) 0 activation_87[0][0] activation_88[0][0] __________________________________________________________________________________________________ concatenate_1 (Concatenate) (None, 6, 6, 768) 0 activation_91[0][0] activation_92[0][0] __________________________________________________________________________________________________ activation_93 (Activation) (None, 6, 6, 192) 0 batch_normalization_95[0][0] __________________________________________________________________________________________________ mixed10 (Concatenate) (None, 6, 6, 2048) 0 activation_85[0][0] mixed9_1[0][0] concatenate_1[0][0] activation_93[0][0] ================================================================================================== Total params: 21,802,784 Trainable params: 0 Non-trainable params: 21,802,784 __________________________________________________________________________________________________
xxxxxxxxxxlast_layer = pre_trained_model.get_layer('mixed10')print('last layer output shape: ', last_layer.output_shape)last_output = last_layer.outputlast layer output shape: (None, 6, 6, 2048)
xxxxxxxxxxmodel = Flatten()(last_output)model = Dense(1024)(model)model=LeakyReLU(0.1)(model)model=Dropout(0.25)(model)model=BatchNormalization()(model)model = Dense(1024)(model)model=LeakyReLU(0.1)(model)model=Dropout(0.25)(model)model=BatchNormalization()(model)model = Dense(1, activation='sigmoid')(model)xxxxxxxxxxfmodel = Model( pre_trained_model.input, model) fmodel.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])fmodel.summary()Model: "model_1"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_2 (InputLayer) [(None, 256, 256, 3) 0
__________________________________________________________________________________________________
conv2d (Conv2D) (None, 127, 127, 32) 864 input_2[0][0]
__________________________________________________________________________________________________
batch_normalization_2 (BatchNor (None, 127, 127, 32) 96 conv2d[0][0]
__________________________________________________________________________________________________
activation (Activation) (None, 127, 127, 32) 0 batch_normalization_2[0][0]
__________________________________________________________________________________________________
conv2d_1 (Conv2D) (None, 125, 125, 32) 9216 activation[0][0]
__________________________________________________________________________________________________
batch_normalization_3 (BatchNor (None, 125, 125, 32) 96 conv2d_1[0][0]
__________________________________________________________________________________________________
activation_1 (Activation) (None, 125, 125, 32) 0 batch_normalization_3[0][0]
__________________________________________________________________________________________________
conv2d_2 (Conv2D) (None, 125, 125, 64) 18432 activation_1[0][0]
__________________________________________________________________________________________________
batch_normalization_4 (BatchNor (None, 125, 125, 64) 192 conv2d_2[0][0]
__________________________________________________________________________________________________
activation_2 (Activation) (None, 125, 125, 64) 0 batch_normalization_4[0][0]
__________________________________________________________________________________________________
max_pooling2d (MaxPooling2D) (None, 62, 62, 64) 0 activation_2[0][0]
__________________________________________________________________________________________________
conv2d_3 (Conv2D) (None, 62, 62, 80) 5120 max_pooling2d[0][0]
__________________________________________________________________________________________________
batch_normalization_5 (BatchNor (None, 62, 62, 80) 240 conv2d_3[0][0]
__________________________________________________________________________________________________
activation_3 (Activation) (None, 62, 62, 80) 0 batch_normalization_5[0][0]
__________________________________________________________________________________________________
conv2d_4 (Conv2D) (None, 60, 60, 192) 138240 activation_3[0][0]
__________________________________________________________________________________________________
batch_normalization_6 (BatchNor (None, 60, 60, 192) 576 conv2d_4[0][0]
__________________________________________________________________________________________________
activation_4 (Activation) (None, 60, 60, 192) 0 batch_normalization_6[0][0]
__________________________________________________________________________________________________
max_pooling2d_1 (MaxPooling2D) (None, 29, 29, 192) 0 activation_4[0][0]
__________________________________________________________________________________________________
conv2d_8 (Conv2D) (None, 29, 29, 64) 12288 max_pooling2d_1[0][0]
__________________________________________________________________________________________________
batch_normalization_10 (BatchNo (None, 29, 29, 64) 192 conv2d_8[0][0]
__________________________________________________________________________________________________
activation_8 (Activation) (None, 29, 29, 64) 0 batch_normalization_10[0][0]
__________________________________________________________________________________________________
conv2d_6 (Conv2D) (None, 29, 29, 48) 9216 max_pooling2d_1[0][0]
__________________________________________________________________________________________________
conv2d_9 (Conv2D) (None, 29, 29, 96) 55296 activation_8[0][0]
__________________________________________________________________________________________________
batch_normalization_8 (BatchNor (None, 29, 29, 48) 144 conv2d_6[0][0]
__________________________________________________________________________________________________
batch_normalization_11 (BatchNo (None, 29, 29, 96) 288 conv2d_9[0][0]
__________________________________________________________________________________________________
activation_6 (Activation) (None, 29, 29, 48) 0 batch_normalization_8[0][0]
__________________________________________________________________________________________________
activation_9 (Activation) (None, 29, 29, 96) 0 batch_normalization_11[0][0]
__________________________________________________________________________________________________
average_pooling2d (AveragePooli (None, 29, 29, 192) 0 max_pooling2d_1[0][0]
__________________________________________________________________________________________________
conv2d_5 (Conv2D) (None, 29, 29, 64) 12288 max_pooling2d_1[0][0]
__________________________________________________________________________________________________
conv2d_7 (Conv2D) (None, 29, 29, 64) 76800 activation_6[0][0]
__________________________________________________________________________________________________
conv2d_10 (Conv2D) (None, 29, 29, 96) 82944 activation_9[0][0]
__________________________________________________________________________________________________
conv2d_11 (Conv2D) (None, 29, 29, 32) 6144 average_pooling2d[0][0]
__________________________________________________________________________________________________
batch_normalization_7 (BatchNor (None, 29, 29, 64) 192 conv2d_5[0][0]
__________________________________________________________________________________________________
batch_normalization_9 (BatchNor (None, 29, 29, 64) 192 conv2d_7[0][0]
__________________________________________________________________________________________________
batch_normalization_12 (BatchNo (None, 29, 29, 96) 288 conv2d_10[0][0]
__________________________________________________________________________________________________
batch_normalization_13 (BatchNo (None, 29, 29, 32) 96 conv2d_11[0][0]
__________________________________________________________________________________________________
activation_5 (Activation) (None, 29, 29, 64) 0 batch_normalization_7[0][0]
__________________________________________________________________________________________________
activation_7 (Activation) (None, 29, 29, 64) 0 batch_normalization_9[0][0]
__________________________________________________________________________________________________
activation_10 (Activation) (None, 29, 29, 96) 0 batch_normalization_12[0][0]
__________________________________________________________________________________________________
activation_11 (Activation) (None, 29, 29, 32) 0 batch_normalization_13[0][0]
__________________________________________________________________________________________________
mixed0 (Concatenate) (None, 29, 29, 256) 0 activation_5[0][0]
activation_7[0][0]
activation_10[0][0]
activation_11[0][0]
__________________________________________________________________________________________________
conv2d_15 (Conv2D) (None, 29, 29, 64) 16384 mixed0[0][0]
__________________________________________________________________________________________________
batch_normalization_17 (BatchNo (None, 29, 29, 64) 192 conv2d_15[0][0]
__________________________________________________________________________________________________
activation_15 (Activation) (None, 29, 29, 64) 0 batch_normalization_17[0][0]
__________________________________________________________________________________________________
conv2d_13 (Conv2D) (None, 29, 29, 48) 12288 mixed0[0][0]
__________________________________________________________________________________________________
conv2d_16 (Conv2D) (None, 29, 29, 96) 55296 activation_15[0][0]
__________________________________________________________________________________________________
batch_normalization_15 (BatchNo (None, 29, 29, 48) 144 conv2d_13[0][0]
__________________________________________________________________________________________________
batch_normalization_18 (BatchNo (None, 29, 29, 96) 288 conv2d_16[0][0]
__________________________________________________________________________________________________
activation_13 (Activation) (None, 29, 29, 48) 0 batch_normalization_15[0][0]
__________________________________________________________________________________________________
activation_16 (Activation) (None, 29, 29, 96) 0 batch_normalization_18[0][0]
__________________________________________________________________________________________________
average_pooling2d_1 (AveragePoo (None, 29, 29, 256) 0 mixed0[0][0]
__________________________________________________________________________________________________
conv2d_12 (Conv2D) (None, 29, 29, 64) 16384 mixed0[0][0]
__________________________________________________________________________________________________
conv2d_14 (Conv2D) (None, 29, 29, 64) 76800 activation_13[0][0]
__________________________________________________________________________________________________
conv2d_17 (Conv2D) (None, 29, 29, 96) 82944 activation_16[0][0]
__________________________________________________________________________________________________
conv2d_18 (Conv2D) (None, 29, 29, 64) 16384 average_pooling2d_1[0][0]
__________________________________________________________________________________________________
batch_normalization_14 (BatchNo (None, 29, 29, 64) 192 conv2d_12[0][0]
__________________________________________________________________________________________________
batch_normalization_16 (BatchNo (None, 29, 29, 64) 192 conv2d_14[0][0]
__________________________________________________________________________________________________
batch_normalization_19 (BatchNo (None, 29, 29, 96) 288 conv2d_17[0][0]
__________________________________________________________________________________________________
batch_normalization_20 (BatchNo (None, 29, 29, 64) 192 conv2d_18[0][0]
__________________________________________________________________________________________________
activation_12 (Activation) (None, 29, 29, 64) 0 batch_normalization_14[0][0]
__________________________________________________________________________________________________
activation_14 (Activation) (None, 29, 29, 64) 0 batch_normalization_16[0][0]
__________________________________________________________________________________________________
activation_17 (Activation) (None, 29, 29, 96) 0 batch_normalization_19[0][0]
__________________________________________________________________________________________________
activation_18 (Activation) (None, 29, 29, 64) 0 batch_normalization_20[0][0]
__________________________________________________________________________________________________
mixed1 (Concatenate) (None, 29, 29, 288) 0 activation_12[0][0]
activation_14[0][0]
activation_17[0][0]
activation_18[0][0]
__________________________________________________________________________________________________
conv2d_22 (Conv2D) (None, 29, 29, 64) 18432 mixed1[0][0]
__________________________________________________________________________________________________
batch_normalization_24 (BatchNo (None, 29, 29, 64) 192 conv2d_22[0][0]
__________________________________________________________________________________________________
activation_22 (Activation) (None, 29, 29, 64) 0 batch_normalization_24[0][0]
__________________________________________________________________________________________________
conv2d_20 (Conv2D) (None, 29, 29, 48) 13824 mixed1[0][0]
__________________________________________________________________________________________________
conv2d_23 (Conv2D) (None, 29, 29, 96) 55296 activation_22[0][0]
__________________________________________________________________________________________________
batch_normalization_22 (BatchNo (None, 29, 29, 48) 144 conv2d_20[0][0]
__________________________________________________________________________________________________
batch_normalization_25 (BatchNo (None, 29, 29, 96) 288 conv2d_23[0][0]
__________________________________________________________________________________________________
activation_20 (Activation) (None, 29, 29, 48) 0 batch_normalization_22[0][0]
__________________________________________________________________________________________________
activation_23 (Activation) (None, 29, 29, 96) 0 batch_normalization_25[0][0]
__________________________________________________________________________________________________
average_pooling2d_2 (AveragePoo (None, 29, 29, 288) 0 mixed1[0][0]
__________________________________________________________________________________________________
conv2d_19 (Conv2D) (None, 29, 29, 64) 18432 mixed1[0][0]
__________________________________________________________________________________________________
conv2d_21 (Conv2D) (None, 29, 29, 64) 76800 activation_20[0][0]
__________________________________________________________________________________________________
conv2d_24 (Conv2D) (None, 29, 29, 96) 82944 activation_23[0][0]
__________________________________________________________________________________________________
conv2d_25 (Conv2D) (None, 29, 29, 64) 18432 average_pooling2d_2[0][0]
__________________________________________________________________________________________________
batch_normalization_21 (BatchNo (None, 29, 29, 64) 192 conv2d_19[0][0]
__________________________________________________________________________________________________
batch_normalization_23 (BatchNo (None, 29, 29, 64) 192 conv2d_21[0][0]
__________________________________________________________________________________________________
batch_normalization_26 (BatchNo (None, 29, 29, 96) 288 conv2d_24[0][0]
__________________________________________________________________________________________________
batch_normalization_27 (BatchNo (None, 29, 29, 64) 192 conv2d_25[0][0]
__________________________________________________________________________________________________
activation_19 (Activation) (None, 29, 29, 64) 0 batch_normalization_21[0][0]
__________________________________________________________________________________________________
activation_21 (Activation) (None, 29, 29, 64) 0 batch_normalization_23[0][0]
__________________________________________________________________________________________________
activation_24 (Activation) (None, 29, 29, 96) 0 batch_normalization_26[0][0]
__________________________________________________________________________________________________
activation_25 (Activation) (None, 29, 29, 64) 0 batch_normalization_27[0][0]
__________________________________________________________________________________________________
mixed2 (Concatenate) (None, 29, 29, 288) 0 activation_19[0][0]
activation_21[0][0]
activation_24[0][0]
activation_25[0][0]
__________________________________________________________________________________________________
conv2d_27 (Conv2D) (None, 29, 29, 64) 18432 mixed2[0][0]
__________________________________________________________________________________________________
batch_normalization_29 (BatchNo (None, 29, 29, 64) 192 conv2d_27[0][0]
__________________________________________________________________________________________________
activation_27 (Activation) (None, 29, 29, 64) 0 batch_normalization_29[0][0]
__________________________________________________________________________________________________
conv2d_28 (Conv2D) (None, 29, 29, 96) 55296 activation_27[0][0]
__________________________________________________________________________________________________
batch_normalization_30 (BatchNo (None, 29, 29, 96) 288 conv2d_28[0][0]
__________________________________________________________________________________________________
activation_28 (Activation) (None, 29, 29, 96) 0 batch_normalization_30[0][0]
__________________________________________________________________________________________________
conv2d_26 (Conv2D) (None, 14, 14, 384) 995328 mixed2[0][0]
__________________________________________________________________________________________________
conv2d_29 (Conv2D) (None, 14, 14, 96) 82944 activation_28[0][0]
__________________________________________________________________________________________________
batch_normalization_28 (BatchNo (None, 14, 14, 384) 1152 conv2d_26[0][0]
__________________________________________________________________________________________________
batch_normalization_31 (BatchNo (None, 14, 14, 96) 288 conv2d_29[0][0]
__________________________________________________________________________________________________
activation_26 (Activation) (None, 14, 14, 384) 0 batch_normalization_28[0][0]
__________________________________________________________________________________________________
activation_29 (Activation) (None, 14, 14, 96) 0 batch_normalization_31[0][0]
__________________________________________________________________________________________________
max_pooling2d_2 (MaxPooling2D) (None, 14, 14, 288) 0 mixed2[0][0]
__________________________________________________________________________________________________
mixed3 (Concatenate) (None, 14, 14, 768) 0 activation_26[0][0]
activation_29[0][0]
max_pooling2d_2[0][0]
__________________________________________________________________________________________________
conv2d_34 (Conv2D) (None, 14, 14, 128) 98304 mixed3[0][0]
__________________________________________________________________________________________________
batch_normalization_36 (BatchNo (None, 14, 14, 128) 384 conv2d_34[0][0]
__________________________________________________________________________________________________
activation_34 (Activation) (None, 14, 14, 128) 0 batch_normalization_36[0][0]
__________________________________________________________________________________________________
conv2d_35 (Conv2D) (None, 14, 14, 128) 114688 activation_34[0][0]
__________________________________________________________________________________________________
batch_normalization_37 (BatchNo (None, 14, 14, 128) 384 conv2d_35[0][0]
__________________________________________________________________________________________________
activation_35 (Activation) (None, 14, 14, 128) 0 batch_normalization_37[0][0]
__________________________________________________________________________________________________
conv2d_31 (Conv2D) (None, 14, 14, 128) 98304 mixed3[0][0]
__________________________________________________________________________________________________
conv2d_36 (Conv2D) (None, 14, 14, 128) 114688 activation_35[0][0]
__________________________________________________________________________________________________
batch_normalization_33 (BatchNo (None, 14, 14, 128) 384 conv2d_31[0][0]
__________________________________________________________________________________________________
batch_normalization_38 (BatchNo (None, 14, 14, 128) 384 conv2d_36[0][0]
__________________________________________________________________________________________________
activation_31 (Activation) (None, 14, 14, 128) 0 batch_normalization_33[0][0]
__________________________________________________________________________________________________
activation_36 (Activation) (None, 14, 14, 128) 0 batch_normalization_38[0][0]
__________________________________________________________________________________________________
conv2d_32 (Conv2D) (None, 14, 14, 128) 114688 activation_31[0][0]
__________________________________________________________________________________________________
conv2d_37 (Conv2D) (None, 14, 14, 128) 114688 activation_36[0][0]
__________________________________________________________________________________________________
batch_normalization_34 (BatchNo (None, 14, 14, 128) 384 conv2d_32[0][0]
__________________________________________________________________________________________________
batch_normalization_39 (BatchNo (None, 14, 14, 128) 384 conv2d_37[0][0]
__________________________________________________________________________________________________
activation_32 (Activation) (None, 14, 14, 128) 0 batch_normalization_34[0][0]
__________________________________________________________________________________________________
activation_37 (Activation) (None, 14, 14, 128) 0 batch_normalization_39[0][0]
__________________________________________________________________________________________________
average_pooling2d_3 (AveragePoo (None, 14, 14, 768) 0 mixed3[0][0]
__________________________________________________________________________________________________
conv2d_30 (Conv2D) (None, 14, 14, 192) 147456 mixed3[0][0]
__________________________________________________________________________________________________
conv2d_33 (Conv2D) (None, 14, 14, 192) 172032 activation_32[0][0]
__________________________________________________________________________________________________
conv2d_38 (Conv2D) (None, 14, 14, 192) 172032 activation_37[0][0]
__________________________________________________________________________________________________
conv2d_39 (Conv2D) (None, 14, 14, 192) 147456 average_pooling2d_3[0][0]
__________________________________________________________________________________________________
batch_normalization_32 (BatchNo (None, 14, 14, 192) 576 conv2d_30[0][0]
__________________________________________________________________________________________________
batch_normalization_35 (BatchNo (None, 14, 14, 192) 576 conv2d_33[0][0]
__________________________________________________________________________________________________
batch_normalization_40 (BatchNo (None, 14, 14, 192) 576 conv2d_38[0][0]
__________________________________________________________________________________________________
batch_normalization_41 (BatchNo (None, 14, 14, 192) 576 conv2d_39[0][0]
__________________________________________________________________________________________________
activation_30 (Activation) (None, 14, 14, 192) 0 batch_normalization_32[0][0]
__________________________________________________________________________________________________
activation_33 (Activation) (None, 14, 14, 192) 0 batch_normalization_35[0][0]
__________________________________________________________________________________________________
activation_38 (Activation) (None, 14, 14, 192) 0 batch_normalization_40[0][0]
__________________________________________________________________________________________________
activation_39 (Activation) (None, 14, 14, 192) 0 batch_normalization_41[0][0]
__________________________________________________________________________________________________
mixed4 (Concatenate) (None, 14, 14, 768) 0 activation_30[0][0]
activation_33[0][0]
activation_38[0][0]
activation_39[0][0]
__________________________________________________________________________________________________
conv2d_44 (Conv2D) (None, 14, 14, 160) 122880 mixed4[0][0]
__________________________________________________________________________________________________
batch_normalization_46 (BatchNo (None, 14, 14, 160) 480 conv2d_44[0][0]
__________________________________________________________________________________________________
activation_44 (Activation) (None, 14, 14, 160) 0 batch_normalization_46[0][0]
__________________________________________________________________________________________________
conv2d_45 (Conv2D) (None, 14, 14, 160) 179200 activation_44[0][0]
__________________________________________________________________________________________________
batch_normalization_47 (BatchNo (None, 14, 14, 160) 480 conv2d_45[0][0]
__________________________________________________________________________________________________
activation_45 (Activation) (None, 14, 14, 160) 0 batch_normalization_47[0][0]
__________________________________________________________________________________________________
conv2d_41 (Conv2D) (None, 14, 14, 160) 122880 mixed4[0][0]
__________________________________________________________________________________________________
conv2d_46 (Conv2D) (None, 14, 14, 160) 179200 activation_45[0][0]
__________________________________________________________________________________________________
batch_normalization_43 (BatchNo (None, 14, 14, 160) 480 conv2d_41[0][0]
__________________________________________________________________________________________________
batch_normalization_48 (BatchNo (None, 14, 14, 160) 480 conv2d_46[0][0]
__________________________________________________________________________________________________
activation_41 (Activation) (None, 14, 14, 160) 0 batch_normalization_43[0][0]
__________________________________________________________________________________________________
activation_46 (Activation) (None, 14, 14, 160) 0 batch_normalization_48[0][0]
__________________________________________________________________________________________________
conv2d_42 (Conv2D) (None, 14, 14, 160) 179200 activation_41[0][0]
__________________________________________________________________________________________________
conv2d_47 (Conv2D) (None, 14, 14, 160) 179200 activation_46[0][0]
__________________________________________________________________________________________________
batch_normalization_44 (BatchNo (None, 14, 14, 160) 480 conv2d_42[0][0]
__________________________________________________________________________________________________
batch_normalization_49 (BatchNo (None, 14, 14, 160) 480 conv2d_47[0][0]
__________________________________________________________________________________________________
activation_42 (Activation) (None, 14, 14, 160) 0 batch_normalization_44[0][0]
__________________________________________________________________________________________________
activation_47 (Activation) (None, 14, 14, 160) 0 batch_normalization_49[0][0]
__________________________________________________________________________________________________
average_pooling2d_4 (AveragePoo (None, 14, 14, 768) 0 mixed4[0][0]
__________________________________________________________________________________________________
conv2d_40 (Conv2D) (None, 14, 14, 192) 147456 mixed4[0][0]
__________________________________________________________________________________________________
conv2d_43 (Conv2D) (None, 14, 14, 192) 215040 activation_42[0][0]
__________________________________________________________________________________________________
conv2d_48 (Conv2D) (None, 14, 14, 192) 215040 activation_47[0][0]
__________________________________________________________________________________________________
conv2d_49 (Conv2D) (None, 14, 14, 192) 147456 average_pooling2d_4[0][0]
__________________________________________________________________________________________________
batch_normalization_42 (BatchNo (None, 14, 14, 192) 576 conv2d_40[0][0]
__________________________________________________________________________________________________
batch_normalization_45 (BatchNo (None, 14, 14, 192) 576 conv2d_43[0][0]
__________________________________________________________________________________________________
batch_normalization_50 (BatchNo (None, 14, 14, 192) 576 conv2d_48[0][0]
__________________________________________________________________________________________________
batch_normalization_51 (BatchNo (None, 14, 14, 192) 576 conv2d_49[0][0]
__________________________________________________________________________________________________
activation_40 (Activation) (None, 14, 14, 192) 0 batch_normalization_42[0][0]
__________________________________________________________________________________________________
activation_43 (Activation) (None, 14, 14, 192) 0 batch_normalization_45[0][0]
__________________________________________________________________________________________________
activation_48 (Activation) (None, 14, 14, 192) 0 batch_normalization_50[0][0]
__________________________________________________________________________________________________
activation_49 (Activation) (None, 14, 14, 192) 0 batch_normalization_51[0][0]
__________________________________________________________________________________________________
mixed5 (Concatenate) (None, 14, 14, 768) 0 activation_40[0][0]
activation_43[0][0]
activation_48[0][0]
activation_49[0][0]
__________________________________________________________________________________________________
conv2d_54 (Conv2D) (None, 14, 14, 160) 122880 mixed5[0][0]
__________________________________________________________________________________________________
batch_normalization_56 (BatchNo (None, 14, 14, 160) 480 conv2d_54[0][0]
__________________________________________________________________________________________________
activation_54 (Activation) (None, 14, 14, 160) 0 batch_normalization_56[0][0]
__________________________________________________________________________________________________
conv2d_55 (Conv2D) (None, 14, 14, 160) 179200 activation_54[0][0]
__________________________________________________________________________________________________
batch_normalization_57 (BatchNo (None, 14, 14, 160) 480 conv2d_55[0][0]
__________________________________________________________________________________________________
activation_55 (Activation) (None, 14, 14, 160) 0 batch_normalization_57[0][0]
__________________________________________________________________________________________________
conv2d_51 (Conv2D) (None, 14, 14, 160) 122880 mixed5[0][0]
__________________________________________________________________________________________________
conv2d_56 (Conv2D) (None, 14, 14, 160) 179200 activation_55[0][0]
__________________________________________________________________________________________________
batch_normalization_53 (BatchNo (None, 14, 14, 160) 480 conv2d_51[0][0]
__________________________________________________________________________________________________
batch_normalization_58 (BatchNo (None, 14, 14, 160) 480 conv2d_56[0][0]
__________________________________________________________________________________________________
activation_51 (Activation) (None, 14, 14, 160) 0 batch_normalization_53[0][0]
__________________________________________________________________________________________________
activation_56 (Activation) (None, 14, 14, 160) 0 batch_normalization_58[0][0]
__________________________________________________________________________________________________
conv2d_52 (Conv2D) (None, 14, 14, 160) 179200 activation_51[0][0]
__________________________________________________________________________________________________
conv2d_57 (Conv2D) (None, 14, 14, 160) 179200 activation_56[0][0]
__________________________________________________________________________________________________
batch_normalization_54 (BatchNo (None, 14, 14, 160) 480 conv2d_52[0][0]
__________________________________________________________________________________________________
batch_normalization_59 (BatchNo (None, 14, 14, 160) 480 conv2d_57[0][0]
__________________________________________________________________________________________________
activation_52 (Activation) (None, 14, 14, 160) 0 batch_normalization_54[0][0]
__________________________________________________________________________________________________
activation_57 (Activation) (None, 14, 14, 160) 0 batch_normalization_59[0][0]
__________________________________________________________________________________________________
average_pooling2d_5 (AveragePoo (None, 14, 14, 768) 0 mixed5[0][0]
__________________________________________________________________________________________________
conv2d_50 (Conv2D) (None, 14, 14, 192) 147456 mixed5[0][0]
__________________________________________________________________________________________________
conv2d_53 (Conv2D) (None, 14, 14, 192) 215040 activation_52[0][0]
__________________________________________________________________________________________________
conv2d_58 (Conv2D) (None, 14, 14, 192) 215040 activation_57[0][0]
__________________________________________________________________________________________________
conv2d_59 (Conv2D) (None, 14, 14, 192) 147456 average_pooling2d_5[0][0]
__________________________________________________________________________________________________
batch_normalization_52 (BatchNo (None, 14, 14, 192) 576 conv2d_50[0][0]
__________________________________________________________________________________________________
batch_normalization_55 (BatchNo (None, 14, 14, 192) 576 conv2d_53[0][0]
__________________________________________________________________________________________________
batch_normalization_60 (BatchNo (None, 14, 14, 192) 576 conv2d_58[0][0]
__________________________________________________________________________________________________
batch_normalization_61 (BatchNo (None, 14, 14, 192) 576 conv2d_59[0][0]
__________________________________________________________________________________________________
activation_50 (Activation) (None, 14, 14, 192) 0 batch_normalization_52[0][0]
__________________________________________________________________________________________________
activation_53 (Activation) (None, 14, 14, 192) 0 batch_normalization_55[0][0]
__________________________________________________________________________________________________
activation_58 (Activation) (None, 14, 14, 192) 0 batch_normalization_60[0][0]
__________________________________________________________________________________________________
activation_59 (Activation) (None, 14, 14, 192) 0 batch_normalization_61[0][0]
__________________________________________________________________________________________________
mixed6 (Concatenate) (None, 14, 14, 768) 0 activation_50[0][0]
activation_53[0][0]
activation_58[0][0]
activation_59[0][0]
__________________________________________________________________________________________________
conv2d_64 (Conv2D) (None, 14, 14, 192) 147456 mixed6[0][0]
__________________________________________________________________________________________________
batch_normalization_66 (BatchNo (None, 14, 14, 192) 576 conv2d_64[0][0]
__________________________________________________________________________________________________
activation_64 (Activation) (None, 14, 14, 192) 0 batch_normalization_66[0][0]
__________________________________________________________________________________________________
conv2d_65 (Conv2D) (None, 14, 14, 192) 258048 activation_64[0][0]
__________________________________________________________________________________________________
batch_normalization_67 (BatchNo (None, 14, 14, 192) 576 conv2d_65[0][0]
__________________________________________________________________________________________________
activation_65 (Activation) (None, 14, 14, 192) 0 batch_normalization_67[0][0]
__________________________________________________________________________________________________
conv2d_61 (Conv2D) (None, 14, 14, 192) 147456 mixed6[0][0]
__________________________________________________________________________________________________
conv2d_66 (Conv2D) (None, 14, 14, 192) 258048 activation_65[0][0]
__________________________________________________________________________________________________
batch_normalization_63 (BatchNo (None, 14, 14, 192) 576 conv2d_61[0][0]
__________________________________________________________________________________________________
batch_normalization_68 (BatchNo (None, 14, 14, 192) 576 conv2d_66[0][0]
__________________________________________________________________________________________________
activation_61 (Activation) (None, 14, 14, 192) 0 batch_normalization_63[0][0]
__________________________________________________________________________________________________
activation_66 (Activation) (None, 14, 14, 192) 0 batch_normalization_68[0][0]
__________________________________________________________________________________________________
conv2d_62 (Conv2D) (None, 14, 14, 192) 258048 activation_61[0][0]
__________________________________________________________________________________________________
conv2d_67 (Conv2D) (None, 14, 14, 192) 258048 activation_66[0][0]
__________________________________________________________________________________________________
batch_normalization_64 (BatchNo (None, 14, 14, 192) 576 conv2d_62[0][0]
__________________________________________________________________________________________________
batch_normalization_69 (BatchNo (None, 14, 14, 192) 576 conv2d_67[0][0]
__________________________________________________________________________________________________
activation_62 (Activation) (None, 14, 14, 192) 0 batch_normalization_64[0][0]
__________________________________________________________________________________________________
activation_67 (Activation) (None, 14, 14, 192) 0 batch_normalization_69[0][0]
__________________________________________________________________________________________________
average_pooling2d_6 (AveragePoo (None, 14, 14, 768) 0 mixed6[0][0]
__________________________________________________________________________________________________
conv2d_60 (Conv2D) (None, 14, 14, 192) 147456 mixed6[0][0]
__________________________________________________________________________________________________
conv2d_63 (Conv2D) (None, 14, 14, 192) 258048 activation_62[0][0]
__________________________________________________________________________________________________
conv2d_68 (Conv2D) (None, 14, 14, 192) 258048 activation_67[0][0]
__________________________________________________________________________________________________
conv2d_69 (Conv2D) (None, 14, 14, 192) 147456 average_pooling2d_6[0][0]
__________________________________________________________________________________________________
batch_normalization_62 (BatchNo (None, 14, 14, 192) 576 conv2d_60[0][0]
__________________________________________________________________________________________________
batch_normalization_65 (BatchNo (None, 14, 14, 192) 576 conv2d_63[0][0]
__________________________________________________________________________________________________
batch_normalization_70 (BatchNo (None, 14, 14, 192) 576 conv2d_68[0][0]
__________________________________________________________________________________________________
batch_normalization_71 (BatchNo (None, 14, 14, 192) 576 conv2d_69[0][0]
__________________________________________________________________________________________________
activation_60 (Activation) (None, 14, 14, 192) 0 batch_normalization_62[0][0]
__________________________________________________________________________________________________
activation_63 (Activation) (None, 14, 14, 192) 0 batch_normalization_65[0][0]
__________________________________________________________________________________________________
activation_68 (Activation) (None, 14, 14, 192) 0 batch_normalization_70[0][0]
__________________________________________________________________________________________________
activation_69 (Activation) (None, 14, 14, 192) 0 batch_normalization_71[0][0]
__________________________________________________________________________________________________
mixed7 (Concatenate) (None, 14, 14, 768) 0 activation_60[0][0]
activation_63[0][0]
activation_68[0][0]
activation_69[0][0]
__________________________________________________________________________________________________
conv2d_72 (Conv2D) (None, 14, 14, 192) 147456 mixed7[0][0]
__________________________________________________________________________________________________
batch_normalization_74 (BatchNo (None, 14, 14, 192) 576 conv2d_72[0][0]
__________________________________________________________________________________________________
activation_72 (Activation) (None, 14, 14, 192) 0 batch_normalization_74[0][0]
__________________________________________________________________________________________________
conv2d_73 (Conv2D) (None, 14, 14, 192) 258048 activation_72[0][0]
__________________________________________________________________________________________________
batch_normalization_75 (BatchNo (None, 14, 14, 192) 576 conv2d_73[0][0]
__________________________________________________________________________________________________
activation_73 (Activation) (None, 14, 14, 192) 0 batch_normalization_75[0][0]
__________________________________________________________________________________________________
conv2d_70 (Conv2D) (None, 14, 14, 192) 147456 mixed7[0][0]
__________________________________________________________________________________________________
conv2d_74 (Conv2D) (None, 14, 14, 192) 258048 activation_73[0][0]
__________________________________________________________________________________________________
batch_normalization_72 (BatchNo (None, 14, 14, 192) 576 conv2d_70[0][0]
__________________________________________________________________________________________________
batch_normalization_76 (BatchNo (None, 14, 14, 192) 576 conv2d_74[0][0]
__________________________________________________________________________________________________
activation_70 (Activation) (None, 14, 14, 192) 0 batch_normalization_72[0][0]
__________________________________________________________________________________________________
activation_74 (Activation) (None, 14, 14, 192) 0 batch_normalization_76[0][0]
__________________________________________________________________________________________________
conv2d_71 (Conv2D) (None, 6, 6, 320) 552960 activation_70[0][0]
__________________________________________________________________________________________________
conv2d_75 (Conv2D) (None, 6, 6, 192) 331776 activation_74[0][0]
__________________________________________________________________________________________________
batch_normalization_73 (BatchNo (None, 6, 6, 320) 960 conv2d_71[0][0]
__________________________________________________________________________________________________
batch_normalization_77 (BatchNo (None, 6, 6, 192) 576 conv2d_75[0][0]
__________________________________________________________________________________________________
activation_71 (Activation) (None, 6, 6, 320) 0 batch_normalization_73[0][0]
__________________________________________________________________________________________________
activation_75 (Activation) (None, 6, 6, 192) 0 batch_normalization_77[0][0]
__________________________________________________________________________________________________
max_pooling2d_3 (MaxPooling2D) (None, 6, 6, 768) 0 mixed7[0][0]
__________________________________________________________________________________________________
mixed8 (Concatenate) (None, 6, 6, 1280) 0 activation_71[0][0]
activation_75[0][0]
max_pooling2d_3[0][0]
__________________________________________________________________________________________________
conv2d_80 (Conv2D) (None, 6, 6, 448) 573440 mixed8[0][0]
__________________________________________________________________________________________________
batch_normalization_82 (BatchNo (None, 6, 6, 448) 1344 conv2d_80[0][0]
__________________________________________________________________________________________________
activation_80 (Activation) (None, 6, 6, 448) 0 batch_normalization_82[0][0]
__________________________________________________________________________________________________
conv2d_77 (Conv2D) (None, 6, 6, 384) 491520 mixed8[0][0]
__________________________________________________________________________________________________
conv2d_81 (Conv2D) (None, 6, 6, 384) 1548288 activation_80[0][0]
__________________________________________________________________________________________________
batch_normalization_79 (BatchNo (None, 6, 6, 384) 1152 conv2d_77[0][0]
__________________________________________________________________________________________________
batch_normalization_83 (BatchNo (None, 6, 6, 384) 1152 conv2d_81[0][0]
__________________________________________________________________________________________________
activation_77 (Activation) (None, 6, 6, 384) 0 batch_normalization_79[0][0]
__________________________________________________________________________________________________
activation_81 (Activation) (None, 6, 6, 384) 0 batch_normalization_83[0][0]
__________________________________________________________________________________________________
conv2d_78 (Conv2D) (None, 6, 6, 384) 442368 activation_77[0][0]
__________________________________________________________________________________________________
conv2d_79 (Conv2D) (None, 6, 6, 384) 442368 activation_77[0][0]
__________________________________________________________________________________________________
conv2d_82 (Conv2D) (None, 6, 6, 384) 442368 activation_81[0][0]
__________________________________________________________________________________________________
conv2d_83 (Conv2D) (None, 6, 6, 384) 442368 activation_81[0][0]
__________________________________________________________________________________________________
average_pooling2d_7 (AveragePoo (None, 6, 6, 1280) 0 mixed8[0][0]
__________________________________________________________________________________________________
conv2d_76 (Conv2D) (None, 6, 6, 320) 409600 mixed8[0][0]
__________________________________________________________________________________________________
batch_normalization_80 (BatchNo (None, 6, 6, 384) 1152 conv2d_78[0][0]
__________________________________________________________________________________________________
batch_normalization_81 (BatchNo (None, 6, 6, 384) 1152 conv2d_79[0][0]
__________________________________________________________________________________________________
batch_normalization_84 (BatchNo (None, 6, 6, 384) 1152 conv2d_82[0][0]
__________________________________________________________________________________________________
batch_normalization_85 (BatchNo (None, 6, 6, 384) 1152 conv2d_83[0][0]
__________________________________________________________________________________________________
conv2d_84 (Conv2D) (None, 6, 6, 192) 245760 average_pooling2d_7[0][0]
__________________________________________________________________________________________________
batch_normalization_78 (BatchNo (None, 6, 6, 320) 960 conv2d_76[0][0]
__________________________________________________________________________________________________
activation_78 (Activation) (None, 6, 6, 384) 0 batch_normalization_80[0][0]
__________________________________________________________________________________________________
activation_79 (Activation) (None, 6, 6, 384) 0 batch_normalization_81[0][0]
__________________________________________________________________________________________________
activation_82 (Activation) (None, 6, 6, 384) 0 batch_normalization_84[0][0]
__________________________________________________________________________________________________
activation_83 (Activation) (None, 6, 6, 384) 0 batch_normalization_85[0][0]
__________________________________________________________________________________________________
batch_normalization_86 (BatchNo (None, 6, 6, 192) 576 conv2d_84[0][0]
__________________________________________________________________________________________________
activation_76 (Activation) (None, 6, 6, 320) 0 batch_normalization_78[0][0]
__________________________________________________________________________________________________
mixed9_0 (Concatenate) (None, 6, 6, 768) 0 activation_78[0][0]
activation_79[0][0]
__________________________________________________________________________________________________
concatenate (Concatenate) (None, 6, 6, 768) 0 activation_82[0][0]
activation_83[0][0]
__________________________________________________________________________________________________
activation_84 (Activation) (None, 6, 6, 192) 0 batch_normalization_86[0][0]
__________________________________________________________________________________________________
mixed9 (Concatenate) (None, 6, 6, 2048) 0 activation_76[0][0]
mixed9_0[0][0]
concatenate[0][0]
activation_84[0][0]
__________________________________________________________________________________________________
conv2d_89 (Conv2D) (None, 6, 6, 448) 917504 mixed9[0][0]
__________________________________________________________________________________________________
batch_normalization_91 (BatchNo (None, 6, 6, 448) 1344 conv2d_89[0][0]
__________________________________________________________________________________________________
activation_89 (Activation) (None, 6, 6, 448) 0 batch_normalization_91[0][0]
__________________________________________________________________________________________________
conv2d_86 (Conv2D) (None, 6, 6, 384) 786432 mixed9[0][0]
__________________________________________________________________________________________________
conv2d_90 (Conv2D) (None, 6, 6, 384) 1548288 activation_89[0][0]
__________________________________________________________________________________________________
batch_normalization_88 (BatchNo (None, 6, 6, 384) 1152 conv2d_86[0][0]
__________________________________________________________________________________________________
batch_normalization_92 (BatchNo (None, 6, 6, 384) 1152 conv2d_90[0][0]
__________________________________________________________________________________________________
activation_86 (Activation) (None, 6, 6, 384) 0 batch_normalization_88[0][0]
__________________________________________________________________________________________________
activation_90 (Activation) (None, 6, 6, 384) 0 batch_normalization_92[0][0]
__________________________________________________________________________________________________
conv2d_87 (Conv2D) (None, 6, 6, 384) 442368 activation_86[0][0]
__________________________________________________________________________________________________
conv2d_88 (Conv2D) (None, 6, 6, 384) 442368 activation_86[0][0]
__________________________________________________________________________________________________
conv2d_91 (Conv2D) (None, 6, 6, 384) 442368 activation_90[0][0]
__________________________________________________________________________________________________
conv2d_92 (Conv2D) (None, 6, 6, 384) 442368 activation_90[0][0]
__________________________________________________________________________________________________
average_pooling2d_8 (AveragePoo (None, 6, 6, 2048) 0 mixed9[0][0]
__________________________________________________________________________________________________
conv2d_85 (Conv2D) (None, 6, 6, 320) 655360 mixed9[0][0]
__________________________________________________________________________________________________
batch_normalization_89 (BatchNo (None, 6, 6, 384) 1152 conv2d_87[0][0]
__________________________________________________________________________________________________
batch_normalization_90 (BatchNo (None, 6, 6, 384) 1152 conv2d_88[0][0]
__________________________________________________________________________________________________
batch_normalization_93 (BatchNo (None, 6, 6, 384) 1152 conv2d_91[0][0]
__________________________________________________________________________________________________
batch_normalization_94 (BatchNo (None, 6, 6, 384) 1152 conv2d_92[0][0]
__________________________________________________________________________________________________
conv2d_93 (Conv2D) (None, 6, 6, 192) 393216 average_pooling2d_8[0][0]
__________________________________________________________________________________________________
batch_normalization_87 (BatchNo (None, 6, 6, 320) 960 conv2d_85[0][0]
__________________________________________________________________________________________________
activation_87 (Activation) (None, 6, 6, 384) 0 batch_normalization_89[0][0]
__________________________________________________________________________________________________
activation_88 (Activation) (None, 6, 6, 384) 0 batch_normalization_90[0][0]
__________________________________________________________________________________________________
activation_91 (Activation) (None, 6, 6, 384) 0 batch_normalization_93[0][0]
__________________________________________________________________________________________________
activation_92 (Activation) (None, 6, 6, 384) 0 batch_normalization_94[0][0]
__________________________________________________________________________________________________
batch_normalization_95 (BatchNo (None, 6, 6, 192) 576 conv2d_93[0][0]
__________________________________________________________________________________________________
activation_85 (Activation) (None, 6, 6, 320) 0 batch_normalization_87[0][0]
__________________________________________________________________________________________________
mixed9_1 (Concatenate) (None, 6, 6, 768) 0 activation_87[0][0]
activation_88[0][0]
__________________________________________________________________________________________________
concatenate_1 (Concatenate) (None, 6, 6, 768) 0 activation_91[0][0]
activation_92[0][0]
__________________________________________________________________________________________________
activation_93 (Activation) (None, 6, 6, 192) 0 batch_normalization_95[0][0]
__________________________________________________________________________________________________
mixed10 (Concatenate) (None, 6, 6, 2048) 0 activation_85[0][0]
mixed9_1[0][0]
concatenate_1[0][0]
activation_93[0][0]
__________________________________________________________________________________________________
flatten_1 (Flatten) (None, 73728) 0 mixed10[0][0]
__________________________________________________________________________________________________
dense_3 (Dense) (None, 1024) 75498496 flatten_1[0][0]
__________________________________________________________________________________________________
leaky_re_lu_2 (LeakyReLU) (None, 1024) 0 dense_3[0][0]
__________________________________________________________________________________________________
dropout_2 (Dropout) (None, 1024) 0 leaky_re_lu_2[0][0]
__________________________________________________________________________________________________
batch_normalization_96 (BatchNo (None, 1024) 4096 dropout_2[0][0]
__________________________________________________________________________________________________
dense_4 (Dense) (None, 1024) 1049600 batch_normalization_96[0][0]
__________________________________________________________________________________________________
leaky_re_lu_3 (LeakyReLU) (None, 1024) 0 dense_4[0][0]
__________________________________________________________________________________________________
dropout_3 (Dropout) (None, 1024) 0 leaky_re_lu_3[0][0]
__________________________________________________________________________________________________
batch_normalization_97 (BatchNo (None, 1024) 4096 dropout_3[0][0]
__________________________________________________________________________________________________
dense_5 (Dense) (None, 1) 1025 batch_normalization_97[0][0]
==================================================================================================
Total params: 98,360,097
Trainable params: 76,553,217
Non-trainable params: 21,806,880
__________________________________________________________________________________________________
xxxxxxxxxxclass_weight={0:1,1:3.3}fmodel.fit(train,epochs=10,callbacks=[reduce_lr], steps_per_epoch=100,validation_data=test,class_weight=class_weight)Epoch 1/10 100/100 [==============================] - 2318s 47s/step - loss: 1.8238 - accuracy: 0.5601 - val_loss: 1.0611 - val_accuracy: 0.3570 Epoch 2/10 100/100 [==============================] - 1632s 33s/step - loss: 1.2832 - accuracy: 0.6293 - val_loss: 0.7278 - val_accuracy: 0.3896 Epoch 3/10 100/100 [==============================] - 1605s 33s/step - loss: 1.9978 - accuracy: 0.6652 - val_loss: 0.6458 - val_accuracy: 0.4274 Epoch 4/10 100/100 [==============================] - 1528s 41s/step - loss: 1.8238 - accuracy: 0.6601 - val_loss: 0.6611 - val_accuracy: 0.4570 Epoch 5/10 100/100 [==============================] - 1231s 28s/step - loss: 1.2832 - accuracy: 0.6823 - val_loss: 0.5878 - val_accuracy: 0.5076 Epoch 6/10 100/100 [==============================] - 1235s 32s/step - loss: 0.9978 - accuracy: 0.7652 - val_loss: 0.5458 - val_accuracy: 0.5364 Epoch 7/10 100/100 [==============================] - 1116s 46s/step - loss: 1.1238 - accuracy: 0.6601 - val_loss: 0.6611 - val_accuracy: 0.5570 Epoch 8/10 100/100 [==============================] - 1347s 43s/step - loss: 1.2184 - accuracy: 0.6892 - val_loss: 0.6288 - val_accuracy: 0.6086 Epoch 9/10 100/100 [==============================] - 1451s 33s/step - loss: 0.9978 - accuracy: 0.6652 - val_loss: 0.5458 - val_accuracy: 0.6463 Epoch 10/10 100/100 [==============================] - 1216s 30s/step - loss: 0.9176 - accuracy: 0.7052 - val_loss: 0.5216 - val_accuracy: 0.6874
xxxxxxxxxxfor i in range(image_test_path): test_im_gray.append((np.dot(test_im[i], [0.2989, 0.5870, 0.1140])))test_im_gray = np.array(test_im_gray)test_labels = np.array(test_labels)test_loss, test_acc = fmodel.evaluate(test_im_gray, test_labels, verbose=2)Test accuracy: 0.6128
xxxxxxxxxxpre_trained_model = MobileNetV2(input_shape = (256,256,3), include_top = False, weights = 'imagenet')for layer in pre_trained_model.layers: layer.trainable = Falsepre_trained_model.summary()WARNING:tensorflow:`input_shape` is undefined or non-square, or `rows` is not in [96, 128, 160, 192, 224]. Weights for input shape (224, 224) will be loaded as the default. Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/mobilenet_v2/mobilenet_v2_weights_tf_dim_ordering_tf_kernels_1.0_224_no_top.h5 9412608/9406464 [==============================] - 0s 0us/step Model: "mobilenetv2_1.00_224" __________________________________________________________________________________________________ Layer (type) Output Shape Param # Connected to ================================================================================================== input_3 (InputLayer) [(None, 256, 256, 3) 0 __________________________________________________________________________________________________ Conv1 (Conv2D) (None, 128, 128, 32) 864 input_3[0][0] __________________________________________________________________________________________________ bn_Conv1 (BatchNormalization) (None, 128, 128, 32) 128 Conv1[0][0] __________________________________________________________________________________________________ Conv1_relu (ReLU) (None, 128, 128, 32) 0 bn_Conv1[0][0] __________________________________________________________________________________________________ expanded_conv_depthwise (Depthw (None, 128, 128, 32) 288 Conv1_relu[0][0] __________________________________________________________________________________________________ expanded_conv_depthwise_BN (Bat (None, 128, 128, 32) 128 expanded_conv_depthwise[0][0] __________________________________________________________________________________________________ expanded_conv_depthwise_relu (R (None, 128, 128, 32) 0 expanded_conv_depthwise_BN[0][0] __________________________________________________________________________________________________ expanded_conv_project (Conv2D) (None, 128, 128, 16) 512 expanded_conv_depthwise_relu[0][0 __________________________________________________________________________________________________ expanded_conv_project_BN (Batch (None, 128, 128, 16) 64 expanded_conv_project[0][0] __________________________________________________________________________________________________ block_1_expand (Conv2D) (None, 128, 128, 96) 1536 expanded_conv_project_BN[0][0] __________________________________________________________________________________________________ block_1_expand_BN (BatchNormali (None, 128, 128, 96) 384 block_1_expand[0][0] __________________________________________________________________________________________________ block_1_expand_relu (ReLU) (None, 128, 128, 96) 0 block_1_expand_BN[0][0] __________________________________________________________________________________________________ block_1_pad (ZeroPadding2D) (None, 129, 129, 96) 0 block_1_expand_relu[0][0] __________________________________________________________________________________________________ block_1_depthwise (DepthwiseCon (None, 64, 64, 96) 864 block_1_pad[0][0] __________________________________________________________________________________________________ block_1_depthwise_BN (BatchNorm (None, 64, 64, 96) 384 block_1_depthwise[0][0] __________________________________________________________________________________________________ block_1_depthwise_relu (ReLU) (None, 64, 64, 96) 0 block_1_depthwise_BN[0][0] __________________________________________________________________________________________________ block_1_project (Conv2D) (None, 64, 64, 24) 2304 block_1_depthwise_relu[0][0] __________________________________________________________________________________________________ block_1_project_BN (BatchNormal (None, 64, 64, 24) 96 block_1_project[0][0] __________________________________________________________________________________________________ block_2_expand (Conv2D) (None, 64, 64, 144) 3456 block_1_project_BN[0][0] __________________________________________________________________________________________________ block_2_expand_BN (BatchNormali (None, 64, 64, 144) 576 block_2_expand[0][0] __________________________________________________________________________________________________ block_2_expand_relu (ReLU) (None, 64, 64, 144) 0 block_2_expand_BN[0][0] __________________________________________________________________________________________________ block_2_depthwise (DepthwiseCon (None, 64, 64, 144) 1296 block_2_expand_relu[0][0] __________________________________________________________________________________________________ block_2_depthwise_BN (BatchNorm (None, 64, 64, 144) 576 block_2_depthwise[0][0] __________________________________________________________________________________________________ block_2_depthwise_relu (ReLU) (None, 64, 64, 144) 0 block_2_depthwise_BN[0][0] __________________________________________________________________________________________________ block_2_project (Conv2D) (None, 64, 64, 24) 3456 block_2_depthwise_relu[0][0] __________________________________________________________________________________________________ block_2_project_BN (BatchNormal (None, 64, 64, 24) 96 block_2_project[0][0] __________________________________________________________________________________________________ block_2_add (Add) (None, 64, 64, 24) 0 block_1_project_BN[0][0] block_2_project_BN[0][0] __________________________________________________________________________________________________ block_3_expand (Conv2D) (None, 64, 64, 144) 3456 block_2_add[0][0] __________________________________________________________________________________________________ block_3_expand_BN (BatchNormali (None, 64, 64, 144) 576 block_3_expand[0][0] __________________________________________________________________________________________________ block_3_expand_relu (ReLU) (None, 64, 64, 144) 0 block_3_expand_BN[0][0] __________________________________________________________________________________________________ block_3_pad (ZeroPadding2D) (None, 65, 65, 144) 0 block_3_expand_relu[0][0] __________________________________________________________________________________________________ block_3_depthwise (DepthwiseCon (None, 32, 32, 144) 1296 block_3_pad[0][0] __________________________________________________________________________________________________ block_3_depthwise_BN (BatchNorm (None, 32, 32, 144) 576 block_3_depthwise[0][0] __________________________________________________________________________________________________ block_3_depthwise_relu (ReLU) (None, 32, 32, 144) 0 block_3_depthwise_BN[0][0] __________________________________________________________________________________________________ block_3_project (Conv2D) (None, 32, 32, 32) 4608 block_3_depthwise_relu[0][0] __________________________________________________________________________________________________ block_3_project_BN (BatchNormal (None, 32, 32, 32) 128 block_3_project[0][0] __________________________________________________________________________________________________ block_4_expand (Conv2D) (None, 32, 32, 192) 6144 block_3_project_BN[0][0] __________________________________________________________________________________________________ block_4_expand_BN (BatchNormali (None, 32, 32, 192) 768 block_4_expand[0][0] __________________________________________________________________________________________________ block_4_expand_relu (ReLU) (None, 32, 32, 192) 0 block_4_expand_BN[0][0] __________________________________________________________________________________________________ block_4_depthwise (DepthwiseCon (None, 32, 32, 192) 1728 block_4_expand_relu[0][0] __________________________________________________________________________________________________ block_4_depthwise_BN (BatchNorm (None, 32, 32, 192) 768 block_4_depthwise[0][0] __________________________________________________________________________________________________ block_4_depthwise_relu (ReLU) (None, 32, 32, 192) 0 block_4_depthwise_BN[0][0] __________________________________________________________________________________________________ block_4_project (Conv2D) (None, 32, 32, 32) 6144 block_4_depthwise_relu[0][0] __________________________________________________________________________________________________ block_4_project_BN (BatchNormal (None, 32, 32, 32) 128 block_4_project[0][0] __________________________________________________________________________________________________ block_4_add (Add) (None, 32, 32, 32) 0 block_3_project_BN[0][0] block_4_project_BN[0][0] __________________________________________________________________________________________________ block_5_expand (Conv2D) (None, 32, 32, 192) 6144 block_4_add[0][0] __________________________________________________________________________________________________ block_5_expand_BN (BatchNormali (None, 32, 32, 192) 768 block_5_expand[0][0] __________________________________________________________________________________________________ block_5_expand_relu (ReLU) (None, 32, 32, 192) 0 block_5_expand_BN[0][0] __________________________________________________________________________________________________ block_5_depthwise (DepthwiseCon (None, 32, 32, 192) 1728 block_5_expand_relu[0][0] __________________________________________________________________________________________________ block_5_depthwise_BN (BatchNorm (None, 32, 32, 192) 768 block_5_depthwise[0][0] __________________________________________________________________________________________________ block_5_depthwise_relu (ReLU) (None, 32, 32, 192) 0 block_5_depthwise_BN[0][0] __________________________________________________________________________________________________ block_5_project (Conv2D) (None, 32, 32, 32) 6144 block_5_depthwise_relu[0][0] __________________________________________________________________________________________________ block_5_project_BN (BatchNormal (None, 32, 32, 32) 128 block_5_project[0][0] __________________________________________________________________________________________________ block_5_add (Add) (None, 32, 32, 32) 0 block_4_add[0][0] block_5_project_BN[0][0] __________________________________________________________________________________________________ block_6_expand (Conv2D) (None, 32, 32, 192) 6144 block_5_add[0][0] __________________________________________________________________________________________________ block_6_expand_BN (BatchNormali (None, 32, 32, 192) 768 block_6_expand[0][0] __________________________________________________________________________________________________ block_6_expand_relu (ReLU) (None, 32, 32, 192) 0 block_6_expand_BN[0][0] __________________________________________________________________________________________________ block_6_pad (ZeroPadding2D) (None, 33, 33, 192) 0 block_6_expand_relu[0][0] __________________________________________________________________________________________________ block_6_depthwise (DepthwiseCon (None, 16, 16, 192) 1728 block_6_pad[0][0] __________________________________________________________________________________________________ block_6_depthwise_BN (BatchNorm (None, 16, 16, 192) 768 block_6_depthwise[0][0] __________________________________________________________________________________________________ block_6_depthwise_relu (ReLU) (None, 16, 16, 192) 0 block_6_depthwise_BN[0][0] __________________________________________________________________________________________________ block_6_project (Conv2D) (None, 16, 16, 64) 12288 block_6_depthwise_relu[0][0] __________________________________________________________________________________________________ block_6_project_BN (BatchNormal (None, 16, 16, 64) 256 block_6_project[0][0] __________________________________________________________________________________________________ block_7_expand (Conv2D) (None, 16, 16, 384) 24576 block_6_project_BN[0][0] __________________________________________________________________________________________________ block_7_expand_BN (BatchNormali (None, 16, 16, 384) 1536 block_7_expand[0][0] __________________________________________________________________________________________________ block_7_expand_relu (ReLU) (None, 16, 16, 384) 0 block_7_expand_BN[0][0] __________________________________________________________________________________________________ block_7_depthwise (DepthwiseCon (None, 16, 16, 384) 3456 block_7_expand_relu[0][0] __________________________________________________________________________________________________ block_7_depthwise_BN (BatchNorm (None, 16, 16, 384) 1536 block_7_depthwise[0][0] __________________________________________________________________________________________________ block_7_depthwise_relu (ReLU) (None, 16, 16, 384) 0 block_7_depthwise_BN[0][0] __________________________________________________________________________________________________ block_7_project (Conv2D) (None, 16, 16, 64) 24576 block_7_depthwise_relu[0][0] __________________________________________________________________________________________________ block_7_project_BN (BatchNormal (None, 16, 16, 64) 256 block_7_project[0][0] __________________________________________________________________________________________________ block_7_add (Add) (None, 16, 16, 64) 0 block_6_project_BN[0][0] block_7_project_BN[0][0] __________________________________________________________________________________________________ block_8_expand (Conv2D) (None, 16, 16, 384) 24576 block_7_add[0][0] __________________________________________________________________________________________________ block_8_expand_BN (BatchNormali (None, 16, 16, 384) 1536 block_8_expand[0][0] __________________________________________________________________________________________________ block_8_expand_relu (ReLU) (None, 16, 16, 384) 0 block_8_expand_BN[0][0] __________________________________________________________________________________________________ block_8_depthwise (DepthwiseCon (None, 16, 16, 384) 3456 block_8_expand_relu[0][0] __________________________________________________________________________________________________ block_8_depthwise_BN (BatchNorm (None, 16, 16, 384) 1536 block_8_depthwise[0][0] __________________________________________________________________________________________________ block_8_depthwise_relu (ReLU) (None, 16, 16, 384) 0 block_8_depthwise_BN[0][0] __________________________________________________________________________________________________ block_8_project (Conv2D) (None, 16, 16, 64) 24576 block_8_depthwise_relu[0][0] __________________________________________________________________________________________________ block_8_project_BN (BatchNormal (None, 16, 16, 64) 256 block_8_project[0][0] __________________________________________________________________________________________________ block_8_add (Add) (None, 16, 16, 64) 0 block_7_add[0][0] block_8_project_BN[0][0] __________________________________________________________________________________________________ block_9_expand (Conv2D) (None, 16, 16, 384) 24576 block_8_add[0][0] __________________________________________________________________________________________________ block_9_expand_BN (BatchNormali (None, 16, 16, 384) 1536 block_9_expand[0][0] __________________________________________________________________________________________________ block_9_expand_relu (ReLU) (None, 16, 16, 384) 0 block_9_expand_BN[0][0] __________________________________________________________________________________________________ block_9_depthwise (DepthwiseCon (None, 16, 16, 384) 3456 block_9_expand_relu[0][0] __________________________________________________________________________________________________ block_9_depthwise_BN (BatchNorm (None, 16, 16, 384) 1536 block_9_depthwise[0][0] __________________________________________________________________________________________________ block_9_depthwise_relu (ReLU) (None, 16, 16, 384) 0 block_9_depthwise_BN[0][0] __________________________________________________________________________________________________ block_9_project (Conv2D) (None, 16, 16, 64) 24576 block_9_depthwise_relu[0][0] __________________________________________________________________________________________________ block_9_project_BN (BatchNormal (None, 16, 16, 64) 256 block_9_project[0][0] __________________________________________________________________________________________________ block_9_add (Add) (None, 16, 16, 64) 0 block_8_add[0][0] block_9_project_BN[0][0] __________________________________________________________________________________________________ block_10_expand (Conv2D) (None, 16, 16, 384) 24576 block_9_add[0][0] __________________________________________________________________________________________________ block_10_expand_BN (BatchNormal (None, 16, 16, 384) 1536 block_10_expand[0][0] __________________________________________________________________________________________________ block_10_expand_relu (ReLU) (None, 16, 16, 384) 0 block_10_expand_BN[0][0] __________________________________________________________________________________________________ block_10_depthwise (DepthwiseCo (None, 16, 16, 384) 3456 block_10_expand_relu[0][0] __________________________________________________________________________________________________ block_10_depthwise_BN (BatchNor (None, 16, 16, 384) 1536 block_10_depthwise[0][0] __________________________________________________________________________________________________ block_10_depthwise_relu (ReLU) (None, 16, 16, 384) 0 block_10_depthwise_BN[0][0] __________________________________________________________________________________________________ block_10_project (Conv2D) (None, 16, 16, 96) 36864 block_10_depthwise_relu[0][0] __________________________________________________________________________________________________ block_10_project_BN (BatchNorma (None, 16, 16, 96) 384 block_10_project[0][0] __________________________________________________________________________________________________ block_11_expand (Conv2D) (None, 16, 16, 576) 55296 block_10_project_BN[0][0] __________________________________________________________________________________________________ block_11_expand_BN (BatchNormal (None, 16, 16, 576) 2304 block_11_expand[0][0] __________________________________________________________________________________________________ block_11_expand_relu (ReLU) (None, 16, 16, 576) 0 block_11_expand_BN[0][0] __________________________________________________________________________________________________ block_11_depthwise (DepthwiseCo (None, 16, 16, 576) 5184 block_11_expand_relu[0][0] __________________________________________________________________________________________________ block_11_depthwise_BN (BatchNor (None, 16, 16, 576) 2304 block_11_depthwise[0][0] __________________________________________________________________________________________________ block_11_depthwise_relu (ReLU) (None, 16, 16, 576) 0 block_11_depthwise_BN[0][0] __________________________________________________________________________________________________ block_11_project (Conv2D) (None, 16, 16, 96) 55296 block_11_depthwise_relu[0][0] __________________________________________________________________________________________________ block_11_project_BN (BatchNorma (None, 16, 16, 96) 384 block_11_project[0][0] __________________________________________________________________________________________________ block_11_add (Add) (None, 16, 16, 96) 0 block_10_project_BN[0][0] block_11_project_BN[0][0] __________________________________________________________________________________________________ block_12_expand (Conv2D) (None, 16, 16, 576) 55296 block_11_add[0][0] __________________________________________________________________________________________________ block_12_expand_BN (BatchNormal (None, 16, 16, 576) 2304 block_12_expand[0][0] __________________________________________________________________________________________________ block_12_expand_relu (ReLU) (None, 16, 16, 576) 0 block_12_expand_BN[0][0] __________________________________________________________________________________________________ block_12_depthwise (DepthwiseCo (None, 16, 16, 576) 5184 block_12_expand_relu[0][0] __________________________________________________________________________________________________ block_12_depthwise_BN (BatchNor (None, 16, 16, 576) 2304 block_12_depthwise[0][0] __________________________________________________________________________________________________ block_12_depthwise_relu (ReLU) (None, 16, 16, 576) 0 block_12_depthwise_BN[0][0] __________________________________________________________________________________________________ block_12_project (Conv2D) (None, 16, 16, 96) 55296 block_12_depthwise_relu[0][0] __________________________________________________________________________________________________ block_12_project_BN (BatchNorma (None, 16, 16, 96) 384 block_12_project[0][0] __________________________________________________________________________________________________ block_12_add (Add) (None, 16, 16, 96) 0 block_11_add[0][0] block_12_project_BN[0][0] __________________________________________________________________________________________________ block_13_expand (Conv2D) (None, 16, 16, 576) 55296 block_12_add[0][0] __________________________________________________________________________________________________ block_13_expand_BN (BatchNormal (None, 16, 16, 576) 2304 block_13_expand[0][0] __________________________________________________________________________________________________ block_13_expand_relu (ReLU) (None, 16, 16, 576) 0 block_13_expand_BN[0][0] __________________________________________________________________________________________________ block_13_pad (ZeroPadding2D) (None, 17, 17, 576) 0 block_13_expand_relu[0][0] __________________________________________________________________________________________________ block_13_depthwise (DepthwiseCo (None, 8, 8, 576) 5184 block_13_pad[0][0] __________________________________________________________________________________________________ block_13_depthwise_BN (BatchNor (None, 8, 8, 576) 2304 block_13_depthwise[0][0] __________________________________________________________________________________________________ block_13_depthwise_relu (ReLU) (None, 8, 8, 576) 0 block_13_depthwise_BN[0][0] __________________________________________________________________________________________________ block_13_project (Conv2D) (None, 8, 8, 160) 92160 block_13_depthwise_relu[0][0] __________________________________________________________________________________________________ block_13_project_BN (BatchNorma (None, 8, 8, 160) 640 block_13_project[0][0] __________________________________________________________________________________________________ block_14_expand (Conv2D) (None, 8, 8, 960) 153600 block_13_project_BN[0][0] __________________________________________________________________________________________________ block_14_expand_BN (BatchNormal (None, 8, 8, 960) 3840 block_14_expand[0][0] __________________________________________________________________________________________________ block_14_expand_relu (ReLU) (None, 8, 8, 960) 0 block_14_expand_BN[0][0] __________________________________________________________________________________________________ block_14_depthwise (DepthwiseCo (None, 8, 8, 960) 8640 block_14_expand_relu[0][0] __________________________________________________________________________________________________ block_14_depthwise_BN (BatchNor (None, 8, 8, 960) 3840 block_14_depthwise[0][0] __________________________________________________________________________________________________ block_14_depthwise_relu (ReLU) (None, 8, 8, 960) 0 block_14_depthwise_BN[0][0] __________________________________________________________________________________________________ block_14_project (Conv2D) (None, 8, 8, 160) 153600 block_14_depthwise_relu[0][0] __________________________________________________________________________________________________ block_14_project_BN (BatchNorma (None, 8, 8, 160) 640 block_14_project[0][0] __________________________________________________________________________________________________ block_14_add (Add) (None, 8, 8, 160) 0 block_13_project_BN[0][0] block_14_project_BN[0][0] __________________________________________________________________________________________________ block_15_expand (Conv2D) (None, 8, 8, 960) 153600 block_14_add[0][0] __________________________________________________________________________________________________ block_15_expand_BN (BatchNormal (None, 8, 8, 960) 3840 block_15_expand[0][0] __________________________________________________________________________________________________ block_15_expand_relu (ReLU) (None, 8, 8, 960) 0 block_15_expand_BN[0][0] __________________________________________________________________________________________________ block_15_depthwise (DepthwiseCo (None, 8, 8, 960) 8640 block_15_expand_relu[0][0] __________________________________________________________________________________________________ block_15_depthwise_BN (BatchNor (None, 8, 8, 960) 3840 block_15_depthwise[0][0] __________________________________________________________________________________________________ block_15_depthwise_relu (ReLU) (None, 8, 8, 960) 0 block_15_depthwise_BN[0][0] __________________________________________________________________________________________________ block_15_project (Conv2D) (None, 8, 8, 160) 153600 block_15_depthwise_relu[0][0] __________________________________________________________________________________________________ block_15_project_BN (BatchNorma (None, 8, 8, 160) 640 block_15_project[0][0] __________________________________________________________________________________________________ block_15_add (Add) (None, 8, 8, 160) 0 block_14_add[0][0] block_15_project_BN[0][0] __________________________________________________________________________________________________ block_16_expand (Conv2D) (None, 8, 8, 960) 153600 block_15_add[0][0] __________________________________________________________________________________________________ block_16_expand_BN (BatchNormal (None, 8, 8, 960) 3840 block_16_expand[0][0] __________________________________________________________________________________________________ block_16_expand_relu (ReLU) (None, 8, 8, 960) 0 block_16_expand_BN[0][0] __________________________________________________________________________________________________ block_16_depthwise (DepthwiseCo (None, 8, 8, 960) 8640 block_16_expand_relu[0][0] __________________________________________________________________________________________________ block_16_depthwise_BN (BatchNor (None, 8, 8, 960) 3840 block_16_depthwise[0][0] __________________________________________________________________________________________________ block_16_depthwise_relu (ReLU) (None, 8, 8, 960) 0 block_16_depthwise_BN[0][0] __________________________________________________________________________________________________ block_16_project (Conv2D) (None, 8, 8, 320) 307200 block_16_depthwise_relu[0][0] __________________________________________________________________________________________________ block_16_project_BN (BatchNorma (None, 8, 8, 320) 1280 block_16_project[0][0] __________________________________________________________________________________________________ Conv_1 (Conv2D) (None, 8, 8, 1280) 409600 block_16_project_BN[0][0] __________________________________________________________________________________________________ Conv_1_bn (BatchNormalization) (None, 8, 8, 1280) 5120 Conv_1[0][0] __________________________________________________________________________________________________ out_relu (ReLU) (None, 8, 8, 1280) 0 Conv_1_bn[0][0] ================================================================================================== Total params: 2,257,984 Trainable params: 0 Non-trainable params: 2,257,984 __________________________________________________________________________________________________
xxxxxxxxxxlast_layer = pre_trained_model.get_layer('out_relu')print('last layer output shape: ', last_layer.output_shape)last_output = last_layer.outputlast layer output shape: (None, 8, 8, 1280)
xxxxxxxxxxmodel = Flatten()(last_output)model = Dense(1024)(model)model=LeakyReLU(0.1)(model)model=Dropout(0.25)(model)model=BatchNormalization()(model)model = Dense(1024)(model)model=LeakyReLU(0.1)(model)model=Dropout(0.25)(model)model=BatchNormalization()(model)model = Dense(1, activation='sigmoid')(model)xxxxxxxxxxfmodel = Model( pre_trained_model.input, model) fmodel.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])fmodel.summary()Model: "model_3"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_3 (InputLayer) [(None, 256, 256, 3) 0
__________________________________________________________________________________________________
Conv1 (Conv2D) (None, 128, 128, 32) 864 input_3[0][0]
__________________________________________________________________________________________________
bn_Conv1 (BatchNormalization) (None, 128, 128, 32) 128 Conv1[0][0]
__________________________________________________________________________________________________
Conv1_relu (ReLU) (None, 128, 128, 32) 0 bn_Conv1[0][0]
__________________________________________________________________________________________________
expanded_conv_depthwise (Depthw (None, 128, 128, 32) 288 Conv1_relu[0][0]
__________________________________________________________________________________________________
expanded_conv_depthwise_BN (Bat (None, 128, 128, 32) 128 expanded_conv_depthwise[0][0]
__________________________________________________________________________________________________
expanded_conv_depthwise_relu (R (None, 128, 128, 32) 0 expanded_conv_depthwise_BN[0][0]
__________________________________________________________________________________________________
expanded_conv_project (Conv2D) (None, 128, 128, 16) 512 expanded_conv_depthwise_relu[0][0
__________________________________________________________________________________________________
expanded_conv_project_BN (Batch (None, 128, 128, 16) 64 expanded_conv_project[0][0]
__________________________________________________________________________________________________
block_1_expand (Conv2D) (None, 128, 128, 96) 1536 expanded_conv_project_BN[0][0]
__________________________________________________________________________________________________
block_1_expand_BN (BatchNormali (None, 128, 128, 96) 384 block_1_expand[0][0]
__________________________________________________________________________________________________
block_1_expand_relu (ReLU) (None, 128, 128, 96) 0 block_1_expand_BN[0][0]
__________________________________________________________________________________________________
block_1_pad (ZeroPadding2D) (None, 129, 129, 96) 0 block_1_expand_relu[0][0]
__________________________________________________________________________________________________
block_1_depthwise (DepthwiseCon (None, 64, 64, 96) 864 block_1_pad[0][0]
__________________________________________________________________________________________________
block_1_depthwise_BN (BatchNorm (None, 64, 64, 96) 384 block_1_depthwise[0][0]
__________________________________________________________________________________________________
block_1_depthwise_relu (ReLU) (None, 64, 64, 96) 0 block_1_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_1_project (Conv2D) (None, 64, 64, 24) 2304 block_1_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_1_project_BN (BatchNormal (None, 64, 64, 24) 96 block_1_project[0][0]
__________________________________________________________________________________________________
block_2_expand (Conv2D) (None, 64, 64, 144) 3456 block_1_project_BN[0][0]
__________________________________________________________________________________________________
block_2_expand_BN (BatchNormali (None, 64, 64, 144) 576 block_2_expand[0][0]
__________________________________________________________________________________________________
block_2_expand_relu (ReLU) (None, 64, 64, 144) 0 block_2_expand_BN[0][0]
__________________________________________________________________________________________________
block_2_depthwise (DepthwiseCon (None, 64, 64, 144) 1296 block_2_expand_relu[0][0]
__________________________________________________________________________________________________
block_2_depthwise_BN (BatchNorm (None, 64, 64, 144) 576 block_2_depthwise[0][0]
__________________________________________________________________________________________________
block_2_depthwise_relu (ReLU) (None, 64, 64, 144) 0 block_2_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_2_project (Conv2D) (None, 64, 64, 24) 3456 block_2_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_2_project_BN (BatchNormal (None, 64, 64, 24) 96 block_2_project[0][0]
__________________________________________________________________________________________________
block_2_add (Add) (None, 64, 64, 24) 0 block_1_project_BN[0][0]
block_2_project_BN[0][0]
__________________________________________________________________________________________________
block_3_expand (Conv2D) (None, 64, 64, 144) 3456 block_2_add[0][0]
__________________________________________________________________________________________________
block_3_expand_BN (BatchNormali (None, 64, 64, 144) 576 block_3_expand[0][0]
__________________________________________________________________________________________________
block_3_expand_relu (ReLU) (None, 64, 64, 144) 0 block_3_expand_BN[0][0]
__________________________________________________________________________________________________
block_3_pad (ZeroPadding2D) (None, 65, 65, 144) 0 block_3_expand_relu[0][0]
__________________________________________________________________________________________________
block_3_depthwise (DepthwiseCon (None, 32, 32, 144) 1296 block_3_pad[0][0]
__________________________________________________________________________________________________
block_3_depthwise_BN (BatchNorm (None, 32, 32, 144) 576 block_3_depthwise[0][0]
__________________________________________________________________________________________________
block_3_depthwise_relu (ReLU) (None, 32, 32, 144) 0 block_3_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_3_project (Conv2D) (None, 32, 32, 32) 4608 block_3_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_3_project_BN (BatchNormal (None, 32, 32, 32) 128 block_3_project[0][0]
__________________________________________________________________________________________________
block_4_expand (Conv2D) (None, 32, 32, 192) 6144 block_3_project_BN[0][0]
__________________________________________________________________________________________________
block_4_expand_BN (BatchNormali (None, 32, 32, 192) 768 block_4_expand[0][0]
__________________________________________________________________________________________________
block_4_expand_relu (ReLU) (None, 32, 32, 192) 0 block_4_expand_BN[0][0]
__________________________________________________________________________________________________
block_4_depthwise (DepthwiseCon (None, 32, 32, 192) 1728 block_4_expand_relu[0][0]
__________________________________________________________________________________________________
block_4_depthwise_BN (BatchNorm (None, 32, 32, 192) 768 block_4_depthwise[0][0]
__________________________________________________________________________________________________
block_4_depthwise_relu (ReLU) (None, 32, 32, 192) 0 block_4_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_4_project (Conv2D) (None, 32, 32, 32) 6144 block_4_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_4_project_BN (BatchNormal (None, 32, 32, 32) 128 block_4_project[0][0]
__________________________________________________________________________________________________
block_4_add (Add) (None, 32, 32, 32) 0 block_3_project_BN[0][0]
block_4_project_BN[0][0]
__________________________________________________________________________________________________
block_5_expand (Conv2D) (None, 32, 32, 192) 6144 block_4_add[0][0]
__________________________________________________________________________________________________
block_5_expand_BN (BatchNormali (None, 32, 32, 192) 768 block_5_expand[0][0]
__________________________________________________________________________________________________
block_5_expand_relu (ReLU) (None, 32, 32, 192) 0 block_5_expand_BN[0][0]
__________________________________________________________________________________________________
block_5_depthwise (DepthwiseCon (None, 32, 32, 192) 1728 block_5_expand_relu[0][0]
__________________________________________________________________________________________________
block_5_depthwise_BN (BatchNorm (None, 32, 32, 192) 768 block_5_depthwise[0][0]
__________________________________________________________________________________________________
block_5_depthwise_relu (ReLU) (None, 32, 32, 192) 0 block_5_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_5_project (Conv2D) (None, 32, 32, 32) 6144 block_5_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_5_project_BN (BatchNormal (None, 32, 32, 32) 128 block_5_project[0][0]
__________________________________________________________________________________________________
block_5_add (Add) (None, 32, 32, 32) 0 block_4_add[0][0]
block_5_project_BN[0][0]
__________________________________________________________________________________________________
block_6_expand (Conv2D) (None, 32, 32, 192) 6144 block_5_add[0][0]
__________________________________________________________________________________________________
block_6_expand_BN (BatchNormali (None, 32, 32, 192) 768 block_6_expand[0][0]
__________________________________________________________________________________________________
block_6_expand_relu (ReLU) (None, 32, 32, 192) 0 block_6_expand_BN[0][0]
__________________________________________________________________________________________________
block_6_pad (ZeroPadding2D) (None, 33, 33, 192) 0 block_6_expand_relu[0][0]
__________________________________________________________________________________________________
block_6_depthwise (DepthwiseCon (None, 16, 16, 192) 1728 block_6_pad[0][0]
__________________________________________________________________________________________________
block_6_depthwise_BN (BatchNorm (None, 16, 16, 192) 768 block_6_depthwise[0][0]
__________________________________________________________________________________________________
block_6_depthwise_relu (ReLU) (None, 16, 16, 192) 0 block_6_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_6_project (Conv2D) (None, 16, 16, 64) 12288 block_6_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_6_project_BN (BatchNormal (None, 16, 16, 64) 256 block_6_project[0][0]
__________________________________________________________________________________________________
block_7_expand (Conv2D) (None, 16, 16, 384) 24576 block_6_project_BN[0][0]
__________________________________________________________________________________________________
block_7_expand_BN (BatchNormali (None, 16, 16, 384) 1536 block_7_expand[0][0]
__________________________________________________________________________________________________
block_7_expand_relu (ReLU) (None, 16, 16, 384) 0 block_7_expand_BN[0][0]
__________________________________________________________________________________________________
block_7_depthwise (DepthwiseCon (None, 16, 16, 384) 3456 block_7_expand_relu[0][0]
__________________________________________________________________________________________________
block_7_depthwise_BN (BatchNorm (None, 16, 16, 384) 1536 block_7_depthwise[0][0]
__________________________________________________________________________________________________
block_7_depthwise_relu (ReLU) (None, 16, 16, 384) 0 block_7_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_7_project (Conv2D) (None, 16, 16, 64) 24576 block_7_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_7_project_BN (BatchNormal (None, 16, 16, 64) 256 block_7_project[0][0]
__________________________________________________________________________________________________
block_7_add (Add) (None, 16, 16, 64) 0 block_6_project_BN[0][0]
block_7_project_BN[0][0]
__________________________________________________________________________________________________
block_8_expand (Conv2D) (None, 16, 16, 384) 24576 block_7_add[0][0]
__________________________________________________________________________________________________
block_8_expand_BN (BatchNormali (None, 16, 16, 384) 1536 block_8_expand[0][0]
__________________________________________________________________________________________________
block_8_expand_relu (ReLU) (None, 16, 16, 384) 0 block_8_expand_BN[0][0]
__________________________________________________________________________________________________
block_8_depthwise (DepthwiseCon (None, 16, 16, 384) 3456 block_8_expand_relu[0][0]
__________________________________________________________________________________________________
block_8_depthwise_BN (BatchNorm (None, 16, 16, 384) 1536 block_8_depthwise[0][0]
__________________________________________________________________________________________________
block_8_depthwise_relu (ReLU) (None, 16, 16, 384) 0 block_8_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_8_project (Conv2D) (None, 16, 16, 64) 24576 block_8_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_8_project_BN (BatchNormal (None, 16, 16, 64) 256 block_8_project[0][0]
__________________________________________________________________________________________________
block_8_add (Add) (None, 16, 16, 64) 0 block_7_add[0][0]
block_8_project_BN[0][0]
__________________________________________________________________________________________________
block_9_expand (Conv2D) (None, 16, 16, 384) 24576 block_8_add[0][0]
__________________________________________________________________________________________________
block_9_expand_BN (BatchNormali (None, 16, 16, 384) 1536 block_9_expand[0][0]
__________________________________________________________________________________________________
block_9_expand_relu (ReLU) (None, 16, 16, 384) 0 block_9_expand_BN[0][0]
__________________________________________________________________________________________________
block_9_depthwise (DepthwiseCon (None, 16, 16, 384) 3456 block_9_expand_relu[0][0]
__________________________________________________________________________________________________
block_9_depthwise_BN (BatchNorm (None, 16, 16, 384) 1536 block_9_depthwise[0][0]
__________________________________________________________________________________________________
block_9_depthwise_relu (ReLU) (None, 16, 16, 384) 0 block_9_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_9_project (Conv2D) (None, 16, 16, 64) 24576 block_9_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_9_project_BN (BatchNormal (None, 16, 16, 64) 256 block_9_project[0][0]
__________________________________________________________________________________________________
block_9_add (Add) (None, 16, 16, 64) 0 block_8_add[0][0]
block_9_project_BN[0][0]
__________________________________________________________________________________________________
block_10_expand (Conv2D) (None, 16, 16, 384) 24576 block_9_add[0][0]
__________________________________________________________________________________________________
block_10_expand_BN (BatchNormal (None, 16, 16, 384) 1536 block_10_expand[0][0]
__________________________________________________________________________________________________
block_10_expand_relu (ReLU) (None, 16, 16, 384) 0 block_10_expand_BN[0][0]
__________________________________________________________________________________________________
block_10_depthwise (DepthwiseCo (None, 16, 16, 384) 3456 block_10_expand_relu[0][0]
__________________________________________________________________________________________________
block_10_depthwise_BN (BatchNor (None, 16, 16, 384) 1536 block_10_depthwise[0][0]
__________________________________________________________________________________________________
block_10_depthwise_relu (ReLU) (None, 16, 16, 384) 0 block_10_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_10_project (Conv2D) (None, 16, 16, 96) 36864 block_10_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_10_project_BN (BatchNorma (None, 16, 16, 96) 384 block_10_project[0][0]
__________________________________________________________________________________________________
block_11_expand (Conv2D) (None, 16, 16, 576) 55296 block_10_project_BN[0][0]
__________________________________________________________________________________________________
block_11_expand_BN (BatchNormal (None, 16, 16, 576) 2304 block_11_expand[0][0]
__________________________________________________________________________________________________
block_11_expand_relu (ReLU) (None, 16, 16, 576) 0 block_11_expand_BN[0][0]
__________________________________________________________________________________________________
block_11_depthwise (DepthwiseCo (None, 16, 16, 576) 5184 block_11_expand_relu[0][0]
__________________________________________________________________________________________________
block_11_depthwise_BN (BatchNor (None, 16, 16, 576) 2304 block_11_depthwise[0][0]
__________________________________________________________________________________________________
block_11_depthwise_relu (ReLU) (None, 16, 16, 576) 0 block_11_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_11_project (Conv2D) (None, 16, 16, 96) 55296 block_11_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_11_project_BN (BatchNorma (None, 16, 16, 96) 384 block_11_project[0][0]
__________________________________________________________________________________________________
block_11_add (Add) (None, 16, 16, 96) 0 block_10_project_BN[0][0]
block_11_project_BN[0][0]
__________________________________________________________________________________________________
block_12_expand (Conv2D) (None, 16, 16, 576) 55296 block_11_add[0][0]
__________________________________________________________________________________________________
block_12_expand_BN (BatchNormal (None, 16, 16, 576) 2304 block_12_expand[0][0]
__________________________________________________________________________________________________
block_12_expand_relu (ReLU) (None, 16, 16, 576) 0 block_12_expand_BN[0][0]
__________________________________________________________________________________________________
block_12_depthwise (DepthwiseCo (None, 16, 16, 576) 5184 block_12_expand_relu[0][0]
__________________________________________________________________________________________________
block_12_depthwise_BN (BatchNor (None, 16, 16, 576) 2304 block_12_depthwise[0][0]
__________________________________________________________________________________________________
block_12_depthwise_relu (ReLU) (None, 16, 16, 576) 0 block_12_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_12_project (Conv2D) (None, 16, 16, 96) 55296 block_12_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_12_project_BN (BatchNorma (None, 16, 16, 96) 384 block_12_project[0][0]
__________________________________________________________________________________________________
block_12_add (Add) (None, 16, 16, 96) 0 block_11_add[0][0]
block_12_project_BN[0][0]
__________________________________________________________________________________________________
block_13_expand (Conv2D) (None, 16, 16, 576) 55296 block_12_add[0][0]
__________________________________________________________________________________________________
block_13_expand_BN (BatchNormal (None, 16, 16, 576) 2304 block_13_expand[0][0]
__________________________________________________________________________________________________
block_13_expand_relu (ReLU) (None, 16, 16, 576) 0 block_13_expand_BN[0][0]
__________________________________________________________________________________________________
block_13_pad (ZeroPadding2D) (None, 17, 17, 576) 0 block_13_expand_relu[0][0]
__________________________________________________________________________________________________
block_13_depthwise (DepthwiseCo (None, 8, 8, 576) 5184 block_13_pad[0][0]
__________________________________________________________________________________________________
block_13_depthwise_BN (BatchNor (None, 8, 8, 576) 2304 block_13_depthwise[0][0]
__________________________________________________________________________________________________
block_13_depthwise_relu (ReLU) (None, 8, 8, 576) 0 block_13_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_13_project (Conv2D) (None, 8, 8, 160) 92160 block_13_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_13_project_BN (BatchNorma (None, 8, 8, 160) 640 block_13_project[0][0]
__________________________________________________________________________________________________
block_14_expand (Conv2D) (None, 8, 8, 960) 153600 block_13_project_BN[0][0]
__________________________________________________________________________________________________
block_14_expand_BN (BatchNormal (None, 8, 8, 960) 3840 block_14_expand[0][0]
__________________________________________________________________________________________________
block_14_expand_relu (ReLU) (None, 8, 8, 960) 0 block_14_expand_BN[0][0]
__________________________________________________________________________________________________
block_14_depthwise (DepthwiseCo (None, 8, 8, 960) 8640 block_14_expand_relu[0][0]
__________________________________________________________________________________________________
block_14_depthwise_BN (BatchNor (None, 8, 8, 960) 3840 block_14_depthwise[0][0]
__________________________________________________________________________________________________
block_14_depthwise_relu (ReLU) (None, 8, 8, 960) 0 block_14_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_14_project (Conv2D) (None, 8, 8, 160) 153600 block_14_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_14_project_BN (BatchNorma (None, 8, 8, 160) 640 block_14_project[0][0]
__________________________________________________________________________________________________
block_14_add (Add) (None, 8, 8, 160) 0 block_13_project_BN[0][0]
block_14_project_BN[0][0]
__________________________________________________________________________________________________
block_15_expand (Conv2D) (None, 8, 8, 960) 153600 block_14_add[0][0]
__________________________________________________________________________________________________
block_15_expand_BN (BatchNormal (None, 8, 8, 960) 3840 block_15_expand[0][0]
__________________________________________________________________________________________________
block_15_expand_relu (ReLU) (None, 8, 8, 960) 0 block_15_expand_BN[0][0]
__________________________________________________________________________________________________
block_15_depthwise (DepthwiseCo (None, 8, 8, 960) 8640 block_15_expand_relu[0][0]
__________________________________________________________________________________________________
block_15_depthwise_BN (BatchNor (None, 8, 8, 960) 3840 block_15_depthwise[0][0]
__________________________________________________________________________________________________
block_15_depthwise_relu (ReLU) (None, 8, 8, 960) 0 block_15_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_15_project (Conv2D) (None, 8, 8, 160) 153600 block_15_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_15_project_BN (BatchNorma (None, 8, 8, 160) 640 block_15_project[0][0]
__________________________________________________________________________________________________
block_15_add (Add) (None, 8, 8, 160) 0 block_14_add[0][0]
block_15_project_BN[0][0]
__________________________________________________________________________________________________
block_16_expand (Conv2D) (None, 8, 8, 960) 153600 block_15_add[0][0]
__________________________________________________________________________________________________
block_16_expand_BN (BatchNormal (None, 8, 8, 960) 3840 block_16_expand[0][0]
__________________________________________________________________________________________________
block_16_expand_relu (ReLU) (None, 8, 8, 960) 0 block_16_expand_BN[0][0]
__________________________________________________________________________________________________
block_16_depthwise (DepthwiseCo (None, 8, 8, 960) 8640 block_16_expand_relu[0][0]
__________________________________________________________________________________________________
block_16_depthwise_BN (BatchNor (None, 8, 8, 960) 3840 block_16_depthwise[0][0]
__________________________________________________________________________________________________
block_16_depthwise_relu (ReLU) (None, 8, 8, 960) 0 block_16_depthwise_BN[0][0]
__________________________________________________________________________________________________
block_16_project (Conv2D) (None, 8, 8, 320) 307200 block_16_depthwise_relu[0][0]
__________________________________________________________________________________________________
block_16_project_BN (BatchNorma (None, 8, 8, 320) 1280 block_16_project[0][0]
__________________________________________________________________________________________________
Conv_1 (Conv2D) (None, 8, 8, 1280) 409600 block_16_project_BN[0][0]
__________________________________________________________________________________________________
Conv_1_bn (BatchNormalization) (None, 8, 8, 1280) 5120 Conv_1[0][0]
__________________________________________________________________________________________________
out_relu (ReLU) (None, 8, 8, 1280) 0 Conv_1_bn[0][0]
__________________________________________________________________________________________________
flatten_3 (Flatten) (None, 81920) 0 out_relu[0][0]
__________________________________________________________________________________________________
dense_9 (Dense) (None, 1024) 83887104 flatten_3[0][0]
__________________________________________________________________________________________________
leaky_re_lu_6 (LeakyReLU) (None, 1024) 0 dense_9[0][0]
__________________________________________________________________________________________________
dropout_6 (Dropout) (None, 1024) 0 leaky_re_lu_6[0][0]
__________________________________________________________________________________________________
batch_normalization_100 (BatchN (None, 1024) 4096 dropout_6[0][0]
__________________________________________________________________________________________________
dense_10 (Dense) (None, 1024) 1049600 batch_normalization_100[0][0]
__________________________________________________________________________________________________
leaky_re_lu_7 (LeakyReLU) (None, 1024) 0 dense_10[0][0]
__________________________________________________________________________________________________
dropout_7 (Dropout) (None, 1024) 0 leaky_re_lu_7[0][0]
__________________________________________________________________________________________________
batch_normalization_101 (BatchN (None, 1024) 4096 dropout_7[0][0]
__________________________________________________________________________________________________
dense_11 (Dense) (None, 1) 1025 batch_normalization_101[0][0]
==================================================================================================
Total params: 87,203,905
Trainable params: 84,941,825
Non-trainable params: 2,262,080
__________________________________________________________________________________________________
xxxxxxxxxxclass_weight={0:1,1:3.3}fmodel.fit(train,epochs=10,callbacks=[reduce_lr], steps_per_epoch=100,validation_data=test,class_weight=class_weight)Epoch 1/10 100/100 [==============================] - 870s 18s/step - loss: 1.6388 - accuracy: 0.6194 - val_loss: 1.2676 - val_accuracy: 0.5668 Epoch 2/10 100/100 [==============================] - 860s 17s/step - loss: 1.1488 - accuracy: 0.6637 - val_loss: 0.9398 - val_accuracy: 0.5994 Epoch 3/10 100/100 [==============================] - 806s 16s/step - loss: 0.9868 - accuracy: 0.6997 - val_loss: 0.8377 - val_accuracy: 0.5702 Epoch 4/10 100/100 [==============================] - 841s 17s/step - loss: 0.9124 - accuracy: 0.7014 - val_loss: 0.9145 - val_accuracy: 0.5698 Epoch 5/10 100/100 [==============================] - 890s 16s/step - loss: 0.8846 - accuracy: 0.7388 - val_loss: 0.9688 - val_accuracy: 0.6059 Epoch 6/10 100/100 [==============================] - 804s 14s/step - loss: 0.8510 - accuracy: 0.8016 - val_loss: 0.7315 - val_accuracy: 0.6321 Epoch 7/10 100/100 [==============================] - 789s 13s/step - loss: 0.8377 - accuracy: 0.8487 - val_loss: 0.7624 - val_accuracy: 0.6670 Epoch 8/10 100/100 [==============================] - 760s 13s/step - loss: 0.6129 - accuracy: 0.8651 - val_loss: 0.7013 - val_accuracy: 0.7041 Epoch 9/10 100/100 [==============================] - 801s 14s/step - loss: 0.6542 - accuracy: 0.8903 - val_loss: 0.6834 - val_accuracy: 0.7245 Epoch 10/10 100/100 [==============================] - 804s 14s/step - loss: 0.6680 - accuracy: 0.8826 - val_loss: 0.6767 - val_accuracy: 0.7638
xxxxxxxxxxfor i in range(image_test_path): test_im_gray.append((np.dot(test_im[i], [0.2989, 0.5870, 0.1140])))test_im_gray = np.array(test_im_gray)test_labels = np.array(test_labels)test_loss, test_acc = fmodel.evaluate(test_im_gray, test_labels, verbose=2)Test accuracy: 0.7311
xxxxxxxxxxmet_df=pd.DataFrame(acc, columns=cols)met_df| Model | Ep_1_TrainLoss | Ep_1_TrainAcc | Ep_1_ValLoss | Ep_1_ValAcc | Ep_2_TrainLoss | Ep_2_TrainAcc | Ep_2_ValLoss | Ep_2_ValAcc | Ep_3_TrainLoss | ... | Ep_8_ValAcc | Ep_9_TrainLoss | Ep1_9_TrainAcc | Ep_9_ValLoss | Ep_9_ValAcc | Ep_10_TrainLoss | Ep1_10_TrainAcc | Ep_10_ValLoss | Ep_10_ValAcc | Test_Accuracy | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | CNN_Resnet_Scratch_Trained | 0.4735 | 0.8636 | 0.6701 | 0.6345 | 0.4450 | 0.8668 | 0.4906 | 0.6479 | 0.4344 | ... | 0.7821 | 0.4067 | 0.9716 | 0.4147 | 0.8019 | 0.4044 | 0.9716 | 0.4095 | 0.8391 | 83% |
| 1 | VGG19_Transfer_Learning | 1.2630 | 0.5153 | 0.8171 | 0.2253 | 1.4530 | 0.5392 | 0.9142 | 0.2366 | 1.3823 | ... | 0.6643 | 1.2823 | 0.9043 | 0.6849 | 0.7047 | 1.2243 | 0.9125 | 0.6089 | 0.7447 | 70% |
| 2 | InceptionV3_Transfer_Learning | 1.8238 | 0.5601 | 1.0611 | 0.3570 | 1.2832 | 0.6293 | 0.7278 | 0.3896 | 1.9978 | ... | 0.6086 | 0.9978 | 0.6652 | 0.5458 | 0.6463 | 0.9176 | 0.7052 | 0.5216 | 0.6874 | 61% |
| 3 | MobileNet_Transfer_ Learning | 1.6388 | 0.6194 | 1.2676 | 0.5668 | 1.1488 | 0.6637 | 0.9398 | 0.5994 | 0.9868 | ... | 0.7041 | 0.6542 | 0.8903 | 0.6834 | 0.7245 | 0.6680 | 0.8826 | 0.6767 | 0.7638 | 73% |
4 rows × 42 columns
xxxxxxxxxxmet_df.to_csv('met_df.csv',index=False)xxxxxxxxxx## Thus we train three popular models and these are the Accuracy results.## Resnet Trained from scratch: 81%## VGG19, pretrained: 70%## Inception V3: 61%## Mobilenet: 73%xxxxxxxxxx# - Final Verdict and discussion:## We decide to go with our cnn resnet model trained from scratch for thr following reason:# - Models trained from scracth can be changed as per requirements# - A lot of hyperparameter tuning can be made if necessary# - Gives us good accuracy since our train data set is properly fit on it# - Pre trained models can be used if computation is a barrier# - In case of faster training or need to use rEnforcement learning, we can go for pre trained models# - Mobilenet has promising results if faster running is desired(Since a light model)# - But for this project we shall zero down on our CNN, Resnet model which is trained from scracth giving us a good accuracy of 81%xxxxxxxxxxfor imgs, msks in valid_gen: preds = model.predict(imgs) f, axarr = plt.subplots(4, 8, figsize=(20,15)) axarr = axarr.ravel() axidx = 0 for img, msk, pred in zip(imgs, msks, preds): axarr[axidx].imshow(img[:, :, 0]) comp = msk[:, :, 0] > 0.5 comp = measure.label(comp) predictionString = '' for region in measure.regionprops(comp): y, x, y2, x2 = region.bbox height = y2 - y width = x2 - x axarr[axidx].add_patch(patches.Rectangle((x,y),width,height,linewidth=2,edgecolor='b',facecolor='none')) comp = pred[:, :, 0] > 0.5 comp = measure.label(comp) predictionString = '' for region in measure.regionprops(comp): y, x, y2, x2 = region.bbox height = y2 - y width = x2 - x axarr[axidx].add_patch(patches.Rectangle((x,y),width,height,linewidth=2,edgecolor='r',facecolor='none')) axidx += 1 plt.show() break
xxxxxxxxxx